Advertising & Public Relations

The big picture

Somewhere today a company will spend two million dollars on a thirty-second commercial, and somewhere else a communications officer will spend an afternoon on a phone call that produces a single newspaper paragraph. Both people believe they are doing the same job: moving what is in an organization's head into yours. They are not doing the same job at all, and the difference between them is the first thing you need to hold clearly, because almost everything else in this course hangs on it.

Advertising is paid, identified persuasion. Someone buys space or time, the sponsor is named, and the sponsor controls what the message says. Public relations is the management of relationships between an organization and the publics that matter to it, and its most distinctive tool is earned attention: coverage, comment, and recommendation that the organization does not pay for and therefore cannot fully control. Advertising buys certainty. Public relations trades certainty away for credibility.

That trade is the axis of this whole field, and you will meet it again in every module. Buy the space and you decide every word, but the audience knows you paid and discounts you accordingly. Earn the coverage and a third party vouches for you, which is worth far more, but the third party may write something you hate. This lesson makes both definitions precise, sets them inside the wider marketing function, and gives you the vocabulary the rest of the course assumes.

Key idea: Advertising buys control over the message; public relations gives control up in exchange for the credibility of a third party.

Advertising, defined carefully

The standard definition has four parts, and each one does work. Advertising is paid, it is non-personal, it is identified, and it is persuasive. Paid separates it from news coverage and from ordinary word of mouth. Non-personal separates it from a salesperson in a showroom, because an advertisement is a single message manufactured once and delivered to many people at once, with no ability to answer a specific person's specific objection. Identified means the sponsor is disclosed, which separates advertising from propaganda and from the disguised material we will spend a whole later lesson on. Persuasive names the intent, which may be to sell, to remind, to reposition, or simply to be remembered.

Notice how much that definition permits. A television spot for a soft drink is advertising. So is a sponsored search result, a billboard, a thirty-second pre-roll video, a direct mail piece, a public service announcement about wildfire prevention, and a full page a company buys to explain its position on a proposed law. The medium is not what makes something advertising. The purchase and the disclosure are.

It is useful to sort advertising by what it is trying to do rather than by where it appears. Product advertising sells a specific good or service. Brand advertising builds the meaning and familiarity of a name over years, expecting no particular sale today. Corporate or institutional advertising promotes the organization itself rather than anything it sells. Advocacy advertising argues a position on a public question. Public service advertising promotes a behavior or cause, usually with donated space, which is how the Ad Council has run campaigns on wildfire prevention, seat belts, and vaccination since 1942. These categories use the same machinery and answer to the same laws.

Key idea: Advertising is paid, non-personal, identified persuasion, and the category covers far more than product commercials, including advocacy and public service work.

Public relations, defined carefully

The Public Relations Society of America defines public relations as a strategic communication process that builds mutually beneficial relationships between organizations and their publics. That definition is deliberately wide, and two of its words repay attention. Strategic means the work serves a stated organizational goal rather than simply generating noise. Relationships, plural and ongoing, means the job is not a single message but a standing connection that has to survive bad news as well as good.

The word publics is the one students most often misread. A public is not the general population. It is any group whose opinion or behavior matters to the organization: customers, employees, investors, regulators, neighbors near a plant, donors, alumni, reporters who cover the sector, activists who oppose it. A hospital's publics include patients, nurses, the state health department, the doctors it wants to recruit, and the neighborhood that objects to its new parking structure. Each of those needs a different message and a different channel, and treating them as one undifferentiated audience is a classic failure.

The practice areas follow from the publics. Media relations works with journalists. Employee or internal communication works with staff. Investor relations works with shareholders and analysts under securities law. Public affairs and government relations work with legislators and regulators. Community relations works with the people who live nearby. Crisis communication is the discipline that runs across all of them when something goes badly wrong. A small organization may have one person doing all of it; a large one has separate teams that occasionally disagree.

Key idea: Public relations manages ongoing relationships with specific publics, and a public is any group whose opinion or behavior matters, not the population at large.

Paid, owned, earned, and shared

The cleanest way to hold the whole communication landscape in your head is the PESO model, which sorts every channel by who owns it and who paid. Paid media is space or time you buy: the commercial, the search ad, the sponsored post, the billboard. Owned media is a channel you control outright: your website, your app, your email list, your store, your annual report, your packaging. Earned media is coverage or comment you did not pay for: a review, a news story, a mention in a podcast, a recommendation from a friend. Shared media is the social layer where owned content circulates and audiences add their own material to it.

Type	Who pays	Control	Credibility	Example
Paid	You	Total	Lowest, audience knows you paid	Television spot, search ad
Owned	You, indirectly	Total	Moderate, clearly your voice	Website, email newsletter
Earned	No one	None	Highest, a third party vouches	News article, product review
Shared	Mixed	Partial	Varies with the sharer	Reposts, comment threads

Read that table as a set of trades rather than a ranking. Paid media is the only kind you can turn on this afternoon at a volume you choose, which is exactly why it exists. Earned media carries more weight per impression, but you cannot schedule it, you cannot guarantee its tone, and a great deal of effort produces nothing at all. Owned media is patient and cheap but reaches only people who already came looking. Mature communication programs run all four at once and know which job each one is doing.

The categories also blur, and where they blur is where the ethics get hard. A paid influencer post looks like shared media. A sponsored article looks like earned media. An advertorial looks like journalism. We will return to each of these under the FTC disclosure rules, because the law's answer is consistent: when paid material is dressed as unpaid material, the audience is being deceived, and disclosure is what fixes it.

Key idea: The PESO model sorts channels into paid, owned, earned, and shared, and every ethical problem in this field appears where paid material is dressed to look unpaid.

Where these functions sit in an organization

Marketing is the larger function that includes advertising. The classic framing is the four Ps: product, price, place, and promotion. Advertising lives inside promotion, alongside sales promotion, personal selling, direct marketing, and public relations. That means an advertising decision is downstream of decisions that matter more. If the product is bad, the price is wrong, or it is not stocked where people shop, no campaign will save it, and the oldest joke in the business is that nothing kills a bad product faster than good advertising, because it gets more people to try it once.

Public relations sits differently. Marketing communication supports selling; corporate communication supports the organization's licence to operate. In many companies public relations reports not to the chief marketing officer but to the chief executive or the general counsel, because when the subject is a product recall, a layoff, a lawsuit, or a regulator's letter, the question is not how to sell more but whether the institution keeps the trust it needs to function. A university, a city government, a hospital, and a charity all run serious public relations operations while doing very little advertising at all.

In practice the two functions overlap constantly and the boundary keeps moving. Integrated marketing communication is the doctrine, now standard, that all of an organization's messages should say a consistent thing across every channel, because audiences do not sort your billboard, your press release, your customer service call, and your chief executive's testimony into separate mental folders. They experience all of it as one organization behaving in one way, and inconsistency reads as dishonesty.

Key idea: Advertising is one tool inside marketing's promotion mix, while public relations also serves the whole organization's standing, which is why it often reports to the chief executive rather than to marketing.

Marketing, propaganda, and the words people use as insults

You will hear public relations called spin and advertising called propaganda, usually by people who mean both as insults. It is worth being exact, because the exact version is more interesting than the insult. Propaganda in its technical sense is communication designed to serve the interests of the communicator, typically without disclosing that purpose and often without regard for accuracy. Advertising discloses its sponsor by definition, so honest advertising is not propaganda; undisclosed persuasion is, and that is precisely why disclosure rules exist.

Spin is a real thing and it is worth naming honestly. Spin is the selective presentation of true facts to produce a misleading overall impression. It is not lying, which is why practitioners can defend it, and it is not honest, which is why audiences resent it. Reporting only the one quarter your sales rose, describing a layoff as a realignment, or announcing bad news at five o'clock on a Friday before a holiday are all spin. None of it is illegal. The PRSA Code of Ethics nonetheless names honesty and disclosure as core obligations, and a practitioner who lives by spin eventually finds that reporters stop returning calls, which destroys the only asset the job actually has.

Here is the standard this course will hold you to, and it is a workable professional standard rather than a purity test. Persuasion is legitimate. Everyone who has ever asked for a raise, argued a case, or written a cover letter has persuaded. What is not legitimate is deceiving the audience about who is speaking, about whether the speaker was paid, or about facts that can be checked. Those three lines will carry you through nearly every ethical question in the field, and we will test them against hard cases in Module 6.

Key idea: Persuasion is legitimate; deception about who is speaking, whether they were paid, or checkable facts is not, and that three-part line resolves most ethical questions in this field.

Common misconceptions

• Public relations is just free advertising. Earned coverage is not free; it costs staff time and relationships, and unlike advertising you cannot control what it says or whether it appears at all.
• A public means the general public. A public is any specific group whose opinion or behavior matters to the organization, and most programs address several at once with different messages.
• Advertising and marketing are the same thing. Advertising is one element of promotion, which is one of marketing's four Ps; product, price, and distribution decisions usually matter more.
• Public relations only matters when something goes wrong. Crisis work is the visible part, but most of the discipline is the ordinary relationship building that determines whether anyone believes you during a crisis.

Recap

• Advertising is paid, non-personal, identified, persuasive communication; the sponsor controls the message.
• Public relations manages ongoing relationships with specific publics and relies on earned attention it cannot control.
• The trade at the center of the field is control versus credibility.
• The PESO model sorts channels into paid, owned, earned, and shared, and ethics problems cluster where paid looks unpaid.
• Advertising sits inside marketing's promotion mix; public relations also serves the organization's standing and often reports higher.
• Legitimate persuasion becomes deception when it hides the speaker, hides payment, or misstates checkable facts.

Sources

1. Public Relations Society of America. (n.d.). About public relations. PRSA. prsa.org
2. Encyclopaedia Britannica. (n.d.). Advertising. britannica.com
3. Encyclopaedia Britannica. (n.d.). Public relations. britannica.com
4. Federal Trade Commission. (n.d.). Advertising and marketing. ftc.gov
5. Ad Council. (n.d.). Our campaigns. adcouncil.org

The big picture

Most histories of advertising are written by the industry, and they read like a parade of geniuses. The real history is more useful and less flattering. Modern advertising was largely invented to sell products that did not work, the regulation that governs it exists because that selling caused real harm, and modern public relations was invented by men working for railroads and tobacco companies who understood the press better than the press understood them. None of that means the field is illegitimate. It means the field's rules were written in blood and money, and knowing why they exist is the difference between following them and understanding them.

We will move through four turning points. The patent medicine era, roughly 1870 to 1906, when national advertising became possible and was immediately abused. The professionalizing decades, when Ivy Lee and Edward Bernays built modern publicity while Congress built the Federal Trade Commission. The Creative Revolution of the late 1950s and 1960s, when the work changed from claims to craft. And the digital turn from the mid 1990s, when the auction replaced the negotiation and the entire economics of the business was rebuilt.

Read this history with one question running underneath it: at each stage, who gained power and who lost it? Advertising's history is not really a history of slogans. It is a history of who could reach a national audience, what that access cost, and who was allowed to check whether the claims were true.

Key idea: The rules governing advertising and public relations were written in response to specific abuses, so learning the history is how you learn why the rules take the shape they do.

Patent medicine and the making of a national market

Three things arrived at roughly the same time in nineteenth century America: railroads that could ship a product nationally, cheap mass-circulation newspapers and magazines, and factories that could make more identical goods than any local market wanted. Together they created the possibility of a national brand, and the first industry to exploit it thoroughly was patent medicine.

The name is a misnomer, since these preparations were rarely patented, which would have required disclosing the formula. They were trademarked instead, and the secrecy was the point. Bottles promising to cure consumption, cancer, catarrh, nervous exhaustion, and infant colic were sold nationally by advertising alone, and many contained substantial alcohol, opium, morphine, or cocaine. Soothing syrups given to teething babies were dosed with morphine. The advertising worked precisely because the claims could not be checked: there was no requirement to list ingredients, no requirement that the claims be true, and no agency with the authority to ask.

Patent medicine was, for decades, the largest category of national advertising in the United States, and it built the machinery everyone else later used. It funded the mass-circulation press. It created the advertising agency, which began as a broker selling newspaper space and evolved into a firm that wrote the copy too. It proved that a manufactured name could be made more valuable than the thing in the bottle. Every technique the field still uses, testimonials, before-and-after images, repetition, the fabricated authority of a doctor's portrait, was tested first on people who were sick.

Key idea: National advertising was built by the patent medicine trade, which proved a brand could be worth more than the product and did so by making claims nobody could check.

The reckoning and the birth of regulation

The correction came from journalism. In 1905 Samuel Hopkins Adams published a series in Collier's Weekly titled The Great American Fraud, naming products, printing analyses of what was actually in them, and documenting deaths. Upton Sinclair's novel The Jungle, published in 1906, did similar work for the meatpacking industry. Public reaction was immediate, and Congress passed the Pure Food and Drug Act in 1906, which required accurate labeling of certain ingredients and created the enforcement body that became the Food and Drug Administration.

Notice what that first law did and did not do. It regulated the label, not the advertisement. A maker could still claim almost anything in a magazine as long as the bottle listed its alcohol and narcotics. Closing that gap took decades. Congress created the Federal Trade Commission in 1914 to police unfair methods of competition, and courts initially read that as protecting competitors rather than consumers. The Wheeler-Lea Act of 1938 fixed the problem directly, giving the FTC explicit authority over unfair and deceptive acts and practices, including false advertising, regardless of whether a competitor was harmed. That amendment is the legal foundation of essentially all modern American advertising regulation.

The industry did not simply wait to be regulated. The advertising clubs, whose descendants include the American Advertising Federation, launched a truth in advertising movement from 1911 onward and set up local vigilance committees to investigate false claims; those committees became the Better Business Bureau. The American Association of Advertising Agencies, the 4A's, was founded in 1917 to set standards for agency conduct and compensation. This pattern, scandal followed by both regulation and self-regulation, repeats through the entire history of the field and is repeating right now around data and disclosure.

Key idea: Muckraking journalism forced the Pure Food and Drug Act in 1906, but advertising itself was not squarely regulated until the Wheeler-Lea Act of 1938 gave the FTC power over deceptive practices.

Ivy Lee, Edward Bernays, and the invention of public relations

Public relations has a founding myth and a founding reality. The myth is that it began as honest counsel. The reality is that it began as damage control for the least popular corporations in America, and that it worked.

Ivy Lee is the first figure who matters. Working for the Pennsylvania Railroad after a 1906 accident, he did something unusual: rather than suppressing the story, he brought reporters to the scene and issued a written statement of facts, arguably the first modern press release. His Declaration of Principles promised that his agency's work would be open, that he would supply prompt and accurate information, and that the sources would always be named. This is genuinely the ethical core of media relations, and practitioners still cite it. It is also true that Lee later worked for John D. Rockefeller Jr. after the 1914 Ludlow Massacre, in which strikers and their families were killed at a Rockefeller-controlled mine, and that critics dubbed him Poison Ivy for the resulting campaign. Both facts belong in the same paragraph.

Edward Bernays is the more consequential and more troubling figure. A nephew of Sigmund Freud, he argued that publics could be moved by appealing to unconscious desires rather than rational argument, and he called the work the engineering of consent. His 1928 book Propaganda opens by asserting that the conscious manipulation of the masses is an important element in democratic society, a sentence worth reading twice. His methods are still the template: rather than advertising bacon, he had a physician survey colleagues about hearty breakfasts and placed the result as news. Rather than advertising cigarettes to women, he arranged for debutantes to light Lucky Strikes in the 1929 Easter parade in New York and briefed the press to describe them as torches of freedom, converting a product into a feminist gesture. He was inventing the third-party endorsement and the manufactured news event at the same time.

Judge that legacy carefully. Bernays proved that publicity could shape behavior at scale, which built the profession. He also worked for the American Tobacco Company in the years when the health evidence was emerging, and in 1954 his firm's work helped the United Fruit Company build American support for the overthrow of Guatemala's elected government. The techniques are neutral; the disclosure and the client are not. Modern codes of ethics, including the PRSA code, exist precisely because Bernays showed what the tools can do when nobody discloses who is holding them.

Key idea: Ivy Lee supplied public relations with its ethic of prompt, named, accurate information, while Bernays supplied its power, and the professional codes exist because his methods worked without disclosure.

Hard sell, then the Creative Revolution

Mid-century American advertising was dominated by the hard sell. Rosser Reeves of the Ted Bates agency codified it as the unique selling proposition: find one specific claim your product can make that competitors cannot, state it plainly, and repeat it without variation until the market has memorized it. Anacin advertising built on this approach ran for years and was widely disliked, and it sold enormous quantities of aspirin. Reeves was explicit that he was not in the entertainment business, and he was right about his own goals.

What broke the hard sell was Bill Bernbach and the agency Doyle Dane Bernbach. Bernbach's structural innovation is easy to miss because it is now universal: he seated the copywriter and the art director together as a team, so that words and image were conceived at once rather than the writer handing copy to a designer to decorate. His creative innovation was to credit the audience with intelligence. The 1959 Volkswagen campaign is the standard example, and it is worth understanding why it was shocking. Postwar American car advertising sold size, chrome, and status. The Volkswagen ads showed a small, plain, German car on a mostly empty white page under the headline Think Small, and the copy admitted the car's limitations in a dry voice that assumed the reader could take a joke.

That campaign changed the industry because it worked commercially, which meant the argument was over. The period from roughly 1959 to 1973 is now called the Creative Revolution: creative departments gained status against account management, agencies hired writers and artists from outside the corporate mold, and the assumption that the audience was a mark to be worn down gave way to the assumption that the audience was a person to be interested. David Ogilvy occupies a middle position in this story, insisting on research and long factual copy while writing with real style, and his dictum that the consumer is not a moron, she is your wife reads as both progressive and dated depending on which half you emphasize.

Key idea: The Creative Revolution replaced repetition of a single claim with craft that credited the audience's intelligence, and it won because campaigns like Volkswagen's sold cars.

The digital turn

The first banner advertisement appeared on HotWired in October 1994, and roughly forty four percent of people who saw it clicked. That number is the single most misleading statistic in the history of the industry, because it set an expectation of measurable response that shaped everything after it. Typical display click rates today are well under one percent, and a large share of clicks are accidental or fraudulent.

Two developments then rebuilt the business. The first was search advertising: Google launched AdWords in 2000 and moved to an auction model in 2002, selling not audiences but intentions, since a person typing a query has already told you what they want. The second was the auction itself, extended in the late 2000s to display inventory through real-time bidding, so that space is no longer negotiated between a media buyer and a publisher's sales representative but sold impression by impression in an automated market that completes in about a tenth of a second. Social platforms then added identity and behavioral data to the same machinery.

Follow the power in that shift and you understand the present. Publishers lost pricing power, because inventory became a commodity sold by machines. Agencies lost their historic monopoly on media access, because a client can buy directly. Advertisers gained targeting precision and lost transparency, since a great deal of money now moves through intermediaries whose fees are hard to see. And audiences became the source of the data that makes the whole system work, which is why privacy law is now advertising law. Every controversy in Module 5, from programmatic fees to attribution to ad fraud, comes out of this rebuilding.

Key idea: The digital turn replaced negotiated media buying with automated auctions, which shifted power away from publishers and agencies toward platforms and made audience data the industry's core asset.

Common misconceptions

• Advertising regulation has always existed. Effective federal authority over deceptive advertising dates only to the Wheeler-Lea Act of 1938; the 1906 food and drug law regulated labels, not advertisements.
• Public relations began as ethical counsel. It began largely as reputation defense for railroads, oil, and tobacco, and its ethical codes were written afterward in response to what its founders proved possible.
• The Creative Revolution was about making ads prettier. It was about a structural change, pairing writer and art director, and a change of assumption about the audience's intelligence, which then produced better commercial results.
• Digital advertising is simply measurable in a way old media was not. Digital produces far more numbers, but as Module 5 shows, most of those numbers do not measure whether the advertising caused anything.

Recap

• National advertising was built by the patent medicine trade on claims nobody could verify.
• Muckraking journalism forced the Pure Food and Drug Act of 1906; the FTC arrived in 1914 and gained real power over deception in 1938.
• The industry answered scandal with self-regulation too, producing the truth in advertising movement, the Better Business Bureau, and the 4A's.
• Ivy Lee gave public relations its ethic of prompt, accurate, attributed information; Bernays gave it its power and its ethical problem.
• The Creative Revolution paired writers with art directors and treated audiences as intelligent, and Volkswagen proved it sold.
• The digital turn replaced negotiation with automated auctions and made audience data the central asset of the business.

Sources

1. Encyclopaedia Britannica. (n.d.). Patent medicine. britannica.com
2. U.S. Food and Drug Administration. (n.d.). FDA history. fda.gov
3. Federal Trade Commission. (n.d.). Our history. ftc.gov
4. Encyclopaedia Britannica. (n.d.). Edward Bernays. britannica.com
5. Encyclopaedia Britannica. (n.d.). David Ogilvy. britannica.com
6. Wikipedia contributors. (n.d.). Doyle Dane Bernbach. Wikipedia. en.wikipedia.org

The big picture

Here is a question that sounds simple and is not. A company sets aside ten million dollars to advertise a product. How much of that money actually buys attention from a human being who might buy the thing? The honest answer, for a digital budget, is that nobody involved can tell you precisely, and the best independent studies suggest a meaningful fraction cannot be traced at all. Understanding why is the most practically useful thing in this lesson, and you cannot understand it without first knowing who all the parties are.

The industry has four principal roles. Clients, properly called advertisers, have something to sell and money to spend. Agencies supply the strategy, the creative work, and the buying expertise the client does not keep in house. Media owners have the audience: broadcasters, publishers, cinema chains, outdoor companies. Platforms are a newer category that behave partly as media owners and partly as marketplaces, and between all of these sit intermediaries, research firms, and trade bodies.

Money flows in one direction through that chain, from advertiser toward media owner, and everybody in between takes something. Where the argument lies is in how much they take and how visibly. Follow the money and you will understand the incentives, and once you understand the incentives, a great deal of otherwise puzzling behavior in this industry becomes obvious.

Key idea: The industry is a chain running from advertiser through agency and intermediaries to media owner, and every current controversy is about who takes what along the way and whether it can be seen.

The client side

Inside a company that advertises, the work usually sits with a marketing department led by a chief marketing officer, with brand managers or product marketing managers owning individual products. A brand manager is a curious role: responsible for a product's performance but usually without direct authority over manufacturing, pricing, or distribution, which means their main lever is communication. That structural fact explains a great deal about why marketing departments believe advertising matters so much.

Communications and public relations often sit in a separate department entirely, reporting to the chief executive or to a chief communications officer, for reasons Lesson 1 explained. In practice the two departments compete for budget and occasionally contradict each other, which is one reason integrated marketing communication became a formal discipline rather than a slogan.

Clients also decide how much work to keep in house. The historical pattern was to outsource almost everything to agencies. Since roughly 2015 many large advertisers have built internal teams, particularly for media buying, social content, and data, driven by cost, speed, and mistrust after a series of transparency disputes. In-housing has real limits: internal teams struggle to attract senior creative talent, and they lose the outside perspective that is arguably the main thing an agency sells. Most large advertisers now run a hybrid.

Key idea: Brand managers own results without owning the product, price, or distribution, which is why communication carries so much weight on the client side, and why in-housing keeps competing with agency work.

Agencies and what is actually inside them

The word agency covers several different businesses. A creative agency develops strategy and makes the work. A media agency plans where the work should run and buys the space. A digital or performance shop specializes in search, social, and measurable response. A public relations firm does earned media and counsel. Specialist shops handle design, direct marketing, experiential events, health care, or multicultural audiences. A full-service agency claims to do most of it.

Inside a creative agency, five functions do the work. Account management owns the client relationship and the schedule, and translates in both directions. Strategy, often called account planning or brand planning, owns the audience and the insight, a role invented in London in the 1960s specifically to represent the consumer inside the agency. The creative department, still organized in the copywriter and art director pairs Bernbach invented, makes the work. Media plans and buys, though at large groups this often sits in a separate company. Production and project management get it built, filmed, cleared, and delivered.

Above the agencies sit holding companies. WPP, Omnicom, Publicis Groupe, Interpublic, and Dentsu own most of the large agency brands you have heard of, along with the media buying arms that place the majority of the world's advertising. In December 2024 Omnicom agreed to acquire Interpublic, which would consolidate the group further. Holding companies matter for a reason students often miss: their scale in media buying gives them purchasing leverage with media owners, and how the benefit of that leverage is shared between the holding company and its clients is precisely the transparency dispute we come to below.

Key idea: An agency is really five functions, account, strategy, creative, media, and production, and most large agencies sit inside a handful of holding companies whose real asset is media buying scale.

How agencies get paid

For most of the twentieth century, agency compensation was simple and strange. The agency bought media on the client's behalf and kept fifteen percent of the gross media cost as commission, a practice inherited from the nineteenth century space brokers. Creative work was effectively free, subsidized by the media commission.

That model has two obvious flaws, and the industry took a long time to admit them. First, it paid the agency more for recommending expensive media, an incentive pointing away from the client's interest. Second, it decoupled payment from effort: a simple campaign in expensive media earned far more than a brilliant campaign in cheap media. From the 1990s onward, most large advertisers moved to alternatives.

Model	How it works	Main weakness
Media commission	Agency keeps a percentage of media spend, historically fifteen percent	Rewards recommending expensive media
Retainer or fee	Client pays for a defined team, usually cost of staff hours plus a margin	Rewards staffing hours, not results; invites scope creep
Project fee	A fixed price for a defined piece of work	Discourages the long relationship strategy needs
Performance or incentive	Part of the fee depends on agreed results	Only as good as the measurement, which Module 5 shows is weak
Value or output pricing	Price per deliverable or per licensed idea	Rare, hard to negotiate, disputed in practice

Public relations firms mostly bill by retainer or by hour, with senior counsel time priced far above junior execution. That has a consequence worth noticing: a PR firm's revenue is tied to hours worked, not to coverage achieved, which is why the pressure to report activity metrics rather than outcomes is chronic, and why the measurement reforms in Module 6 matter.

Key idea: The old fifteen percent media commission paid agencies more for spending more, and the fee, project, and performance models that replaced it each carry their own distortion.

Where the money actually goes in digital media

Now return to the opening question. When a digital display budget moves from an advertiser to a publisher, it usually passes through a demand-side platform that buys on the advertiser's behalf, an exchange that runs the auction, a supply-side platform that represents the publisher, and often verification, data, and identity vendors that each take a fee. Each takes a percentage. This stack is sometimes called the ad tech tax.

Two independent studies made the scale of this concrete. A study commissioned by ISBA and conducted by PwC, published in 2020, traced individual impressions from fifteen advertisers to twelve publishers and found that roughly half of advertiser spend reached publishers, while about fifteen percent of the money could not be attributed to any identified party at all. The researchers called it an unknown delta. In 2023 the Association of National Advertisers published a programmatic transparency study reporting similar structural problems, including large sums flowing to low-quality made-for-advertising sites.

Three further leaks matter. Fraud: bots that generate impressions and clicks no human ever saw, which is why Uber's 2017 lawsuit against its mobile agency and the wider ad fraud literature belong in this lesson rather than only in a chapter on ethics. Viewability: an impression counted and paid for that never entered a human's screen. And principal or inventory media, where an agency group buys media in bulk as principal and resells it to clients at a marked-up price rather than acting purely as the client's agent, a practice that is legal when disclosed and was the center of a 2016 ANA report on media transparency.

Key idea: Independent audits have found that only around half of programmatic display spend reaches the publisher, with fees, fraud, unviewable impressions, and undisclosed principal trading absorbing the rest.

The rest of the ecosystem

Two further groups complete the map. Measurement and research firms sell the numbers everyone argues about: Nielsen for television and audio audiences, Comscore for digital, plus panel providers, brand tracking firms, and the platforms' own reporting, which is not independent and should never be treated as if it were. When a platform grades its own homework, treat the grade as a marketing claim.

Trade and professional bodies set standards and lobby. The Association of National Advertisers represents advertisers, the American Association of Advertising Agencies represents agencies, the Interactive Advertising Bureau represents digital media and sets the technical standards that make programmatic work, the American Advertising Federation runs the industry's broad membership and its awards, and the Public Relations Society of America maintains the profession's code of ethics and accreditation. Self-regulation runs through BBB National Programs, whose National Advertising Division hears challenges to advertising claims, usually brought by competitors, and whose Children's Advertising Review Unit polices advertising to children.

Understand why that self-regulatory layer exists. The FTC has limited resources and cannot review the millions of claims made every day. A competitor, however, has both the motive and the expertise to notice a false claim immediately. Referral to the FTC after a company refuses to comply with a National Advertising Division decision gives the system teeth without requiring the government to do the initial work. It is imperfect, industry funded, and considerably better than nothing.

Key idea: Independent measurement firms, trade bodies, and the self-regulatory system fill gaps the government cannot cover, but platform-supplied numbers are marketing claims, not measurement.

Common misconceptions

• Agencies make their money on creative work. Historically creative was subsidized by a fifteen percent media commission, and even today media buying, data, and production carry much of the margin.
• Programmatic buying is cheaper because it removes middlemen. It replaced one negotiated relationship with a stack of automated intermediaries, each taking a percentage.
• Platform-reported campaign results are measurement. A platform reporting on its own effectiveness has an obvious interest; independent verification exists for exactly this reason.
• In-housing always saves money. It can, but internal teams struggle to attract senior creative talent and lose the outside perspective that is much of what an agency actually sells.

Recap

• The chain runs advertiser, agency, intermediaries, media owner, with platforms straddling the last two roles.
• Agencies contain five functions: account, strategy, creative, media, and production, mostly owned by a few holding companies.
• The fifteen percent media commission gave way to fees, project prices, and performance deals, each with its own distortion.
• The ISBA and PwC study found roughly half of programmatic spend reaching publishers and about fifteen percent untraceable.
• Fraud, unviewable impressions, and undisclosed principal media trading are the other main leaks.
• Trade bodies and BBB National Programs supply a self-regulatory layer that backstops limited FTC resources.

Sources

1. Association of National Advertisers. (n.d.). Industry initiatives and research. ANA. ana.net
2. American Association of Advertising Agencies. (n.d.). About the 4A's. aaaa.org
3. Interactive Advertising Bureau. (n.d.). Standards and guidelines. IAB. iab.com
4. BBB National Programs. (n.d.). National Advertising Division and CARU. bbbprograms.org
5. Wikipedia contributors. (n.d.). Ad fraud. Wikipedia. en.wikipedia.org

The big picture

An agency once opened a presentation with the finding that seventy eight percent of the target audience said convenience was important to them. That is a fact. It is also completely useless, because it is true of nearly every audience for nearly every product, and no creative team on earth can make anything out of it. Two weeks later the same team came back with a different sentence: people are not short of time, they are short of the feeling that they are on top of things. That one could be worked with. The difference between those two sentences is the difference between a finding and an insight, and learning to tell them apart is the whole of this lesson.

Strategy in this field means choosing. Choosing which people to talk to, choosing what to say to them, and, hardest of all, choosing what not to say. Research is what makes those choices defensible rather than arbitrary. Done badly, research is a ritual that produces a deck nobody reads and gives everyone cover. Done well, it produces one sentence about human behavior that makes an expensive problem suddenly tractable.

This lesson covers three things in order: what the different research methods can and cannot tell you, why asking people about their own motives is far less reliable than beginners assume, and how to divide a market and pick a piece of it to pursue. Along the way I will be blunt about the failure modes, because in commercial research the pressure to produce a confident answer is intense and the incentive to admit uncertainty is close to zero.

Key idea: A finding describes what people do; an insight explains something about why, in a way specific enough that a creative team can build on it.

The research toolkit

Start with the simplest distinction. Secondary research is work someone else already did: census data, industry reports, published academic studies, government statistics, the client's own sales history. Primary research is work you commission yourself. The professional habit worth building is to exhaust the secondary sources first, because they are cheaper, faster, and often better than what a rushed primary study will produce, and because it is embarrassing to spend forty thousand dollars discovering something the Bureau of Labor Statistics publishes for free.

Within primary research, the split is qualitative and quantitative, and they answer different questions. Qualitative methods, including depth interviews, focus groups, ethnography, and diary studies, work with small numbers of people in detail. They are good at generating hypotheses, surfacing language people actually use, and revealing motives nobody thought to ask about. They cannot tell you how common anything is. Quantitative methods, including surveys, panels, purchase data, and controlled experiments, work with large numbers. They can tell you how common something is and, when properly designed as an experiment, whether one thing caused another. They cannot tell you about anything you did not think to ask.

A serious study uses both in sequence: qualitative work to find out what the right questions are, quantitative work to find out how many people the answers apply to. Reversing that order, or skipping the first half, produces surveys full of questions that made sense to the client and to nobody else. Focus groups deserve one specific warning. Put eight strangers in a room and you get group dynamics, not eight independent opinions: one confident person anchors the discussion, agreement is socially rewarded, and the moderator's phrasing shapes the answer. They remain useful for hearing how people talk. They are close to worthless for predicting what people will buy.

Key idea: Qualitative research finds out what the questions are, quantitative research finds out how many people the answers cover, and running them in the wrong order wastes both.

Why asking people why is unreliable

Beginners assume that if you want to know why someone bought something, you ask them. The evidence says the answer you get will be fluent, confident, and frequently wrong, and the reason is not that people lie. In a classic 1977 review, Richard Nisbett and Timothy Wilson gathered evidence that people often have no direct access to the cognitive processes producing their own judgments, and that when asked to explain themselves they generate a plausible theory rather than reporting an observation. Their title says it: telling more than we can know.

Stack three further problems on top of that. Social desirability bias means people report the behavior they would like to be seen having: more exercise, less television, more concern for the environment than their purchases show. Rationalization after the fact means a shopper who chose on impulse will supply reasons on request, and will believe them. And demand characteristics mean respondents work out what the researcher seems to want and helpfully provide it, particularly in a room with a moderator and a two-way mirror.

The practical response is to weight revealed preference above stated preference wherever you can get it. What people actually bought, clicked, kept, returned, or renewed is evidence. What they say they would do is a hypothesis. New Coke is the standard teaching case here and it is worth getting right: the taste tests were competently run and sincerely answered, and people did prefer the sweeter formula in a sip test. What the research did not ask, because nobody thought to, was how people would feel about losing the original, and the sip test was a poor model of drinking a whole can. The failure was not dishonest respondents. It was a well-executed study answering a question that was not the real one.

Key idea: People confidently explain motives they cannot actually observe, so revealed behavior outranks stated preference, and a competent study can still fail by asking the wrong question.

From findings to insight

An insight is not a statistic and it is not a description of the target. It is a statement about human behavior that is true, specific, non-obvious, and useful, and all four words are load bearing. True means the evidence supports it. Specific means it applies to this category and these people rather than to humanity in general. Non-obvious means it makes someone in the room sit up slightly. Useful means a creative team can do something with it tomorrow.

Test yourself with three sentences about a bank. First: customers value trust and security. True, entirely obvious, useless, and true of every bank. Second: forty two percent of customers under thirty check their balance more than once a day. True, specific, and still just a finding. Third: people check their balance compulsively not because they have forgotten the number but because they are looking for reassurance that nothing has gone wrong without them noticing. That is an insight, and a campaign or a product feature can be built on it directly.

Getting there is mostly an act of interpretation rather than of data collection, and it is where planners earn their keep. The reliable technique is to look for tension: a gap between what people say and what they do, between what the category assumes and what customers experience, or between two things customers want that cannot both be had. Advertising that feels true almost always sits on top of a tension somebody noticed and named.

Key idea: An insight is true, specific, non-obvious, and useful, and it usually comes from naming a tension between what people say and what they actually do.

Segmentation

Segmentation divides a market into groups that behave differently enough to be worth treating differently. There are four common bases, and mature strategies usually combine them. Demographic segmentation uses age, income, household composition, education. Geographic uses location, climate, and urban or rural setting. Psychographic uses values, attitudes, interests, and lifestyle. Behavioral uses what people actually do: purchase frequency, occasion of use, loyalty, benefits sought.

Basis	Example variable	Strength	Weakness
Demographic	Age, income, household size	Easy to measure and buy media against	Weak predictor of what people actually want
Geographic	Region, climate, urban density	Directly actionable for distribution	Often irrelevant to motive
Psychographic	Values, attitudes, lifestyle	Closer to motivation	Hard to measure reliably or target
Behavioral	Usage occasion, benefit sought, loyalty	Closest to the actual decision	Needs real behavioral data

Behavioral segmentation deserves the most attention because it is closest to the decision being made. Segmenting by occasion is particularly powerful: the same person buys coffee for very different reasons at seven in the morning and at three in the afternoon, and treating them as one coffee drinker loses both. This is the practical version of what the Ehrenberg-Bass Institute calls category entry points, the situations that trigger a purchase, and it explains why a brand often gains more by owning more occasions than by owning more of a demographic.

Key idea: Demographic segments are easy to buy media against but weak at predicting behavior, while behavioral and occasion-based segments sit much closer to the actual decision.

Targeting, and the case against targeting too tightly

Having divided the market, you choose. A usable segment must meet five tests: it must be measurable, so you can size it; substantial, so it is worth the effort; accessible, so you can actually reach it; differentiable, so it responds to your offer differently than other segments; and actionable, so you can serve it with the resources you have. A segment that fails any of these is an interesting observation rather than a target.

Then comes the part that surprises students, because it contradicts the industry's instinct. Digital targeting made it possible to talk to a very narrow audience cheaply, and much of the industry concluded that narrower is always better. The evidence from the Ehrenberg-Bass Institute, associated with Byron Sharp's work, points the other way for most established brands. Brands generally grow by increasing penetration, that is, by being bought occasionally by many light buyers, rather than by extracting more from a loyal few. Light and non-buyers are numerous, and they are precisely the people a tightly targeted campaign excludes. The related finding, sometimes called double jeopardy, is that smaller brands suffer twice: they have fewer buyers and those buyers are somewhat less loyal.

Hold both ideas at once, because the resolution is practical rather than ideological. Narrow targeting is right when your product genuinely serves a narrow need, when the purchase is rare and expensive, or when you are running a direct response campaign to people already in the market. Broad reach is right when you are building a brand that has to be remembered months before anyone needs it. The common error is using precision targeting, which is cheap and produces satisfying dashboards, for a job that actually required reach.

Key idea: Targeting must be measurable, substantial, accessible, differentiable, and actionable, but for established brands the evidence favors broad reach over narrow precision, because growth comes mainly from light buyers.

Common misconceptions

• If you want to know why people buy, ask them. People generate plausible theories about their own behavior rather than reporting it, so revealed behavior outranks stated reasons.
• Focus groups predict what will sell. They surface language and hypotheses; group dynamics make them poor at prediction, and eight people are not a sample.
• An insight is a striking statistic. A statistic is a finding; an insight explains something about behavior specifically enough to act on.
• Narrower targeting is always more efficient. For established brands, growth comes mostly from many light buyers, and tight targeting systematically excludes them.

Recap

• Exhaust secondary research before commissioning primary work.
• Qualitative research generates hypotheses; quantitative research sizes them and, when experimental, tests causation.
• Nisbett and Wilson showed people often cannot observe their own reasons, which is why stated preference is weak evidence.
• New Coke failed not through bad execution but by asking a question that was not the real one.
• An insight is true, specific, non-obvious, and useful, and usually names a tension.
• Segments must be measurable, substantial, accessible, differentiable, and actionable; for most brands, reach beats precision.

Sources

1. Nisbett, R. E., and Wilson, T. D. (1977). Telling more than we can know: Verbal reports on mental processes. Psychological Review, 84(3), 231-259. doi.org
2. Ehrenberg-Bass Institute for Marketing Science. (n.d.). Marketing science research. marketingscience.info
3. Wikipedia contributors. (n.d.). Double jeopardy (marketing). Wikipedia. en.wikipedia.org
4. Encyclopaedia Britannica. (n.d.). Marketing. britannica.com
5. Pew Research Center. (n.d.). Internet and technology research. pewresearch.org

The big picture

You now have an audience and an insight. What you do not yet have is a decision about what your brand is for and a document a creative team can work from. Those two things, positioning and the brief, are where most of the value in strategy actually sits, and where most of the failure happens. A weak brief is not a paperwork problem. It is a client and an agency agreeing not to make a decision, and then paying creative people to make it for them by accident.

Al Ries and Jack Trout put the underlying idea plainly in the 1970s: positioning is not what you do to a product, it is what you do to the mind of the prospect. You cannot rearrange the market, but you can decide which small space in someone's head you are trying to occupy, and you can be consistent about it for long enough that it takes. The discipline is entirely about subtraction. Anyone can list ten good things about a product. Positioning is choosing the one you will be known for and accepting the nine you will not.

This lesson works in three moves. First, positioning: how to define a competitive space and write it down. Second, the creative brief: the standard fields, what a single-minded proposition really means, and the specific ways briefs go wrong. Third, brand equity: what a brand is actually worth, and the sharp disagreement between the two main schools about what builds it.

Key idea: Positioning is a decision about what to give up, and the brief is the document that forces that decision into a form a creative team can act on.

Positioning

A position is defined relative to competitors, never in isolation. The two building blocks are points of parity and points of difference. Points of parity are the things you must be credible on simply to be considered: a bank must be secure, an airline must be safe, a toothpaste must clean teeth. You do not win on these, but you are disqualified without them. Points of difference are the associations you hold that competitors do not, and they are what the position is actually made of.

Beginners spend all their energy on difference and neglect parity, which produces the classic failure of a challenger brand that is genuinely distinctive and that nobody trusts to do the basic job. Establish parity, then compete on difference. Volvo is the durable teaching example: safety was the point of difference, but the cars also had to be recognizably competent on comfort, price, and style, or safety alone would have read as a consolation prize.

A perceptual map is the standard tool for seeing this. Draw two axes representing the attributes that actually drive choice in the category, plot every competitor as customers perceive them rather than as their marketing claims, and look for the empty space. Two warnings. Empty space is sometimes empty because nobody wants what is there, so an unoccupied corner is a hypothesis and not a strategy. And the axes must reflect perception, which means they come from research; a map drawn from the client's own view of the category simply reproduces the client's blind spots.

A positioning statement is the written result, and the conventional template is deliberately rigid: for a specified target, brand X is the frame of reference that delivers a stated benefit because of a stated reason to believe. Fill that in for a real brand you know and notice how uncomfortable the sentence becomes as soon as you have to name a single benefit. The discomfort is the tool working.

Key idea: A position is built from points of parity you must have to compete and one point of difference you will be known for, defined by how customers perceive the category rather than how you describe it.

The creative brief

The brief is the bridge from strategy to work. Formats differ across agencies, but nearly every good brief answers the same questions, and if you can answer these eight honestly you have a brief.

Field	The question it answers	Common failure
Business problem	What commercial situation are we trying to change?	Stated as a communication task instead of a business one
Objective	What must the communication achieve, measurably?	Increase awareness, with no number and no timeframe
Audience	Who exactly are we talking to, and what are they like?	A demographic range with no human detail
Insight	What do we understand about them that others do not?	A finding in an insight's clothing
Proposition	What is the single thing we want them to take out?	Three things joined by commas
Reason to believe	Why should they believe it?	Missing entirely, so the claim floats
Tone and brand assets	How should it feel, and what must be recognizable?	A list of adjectives that could describe anyone
Mandatories	What must appear, legally or practically?	A list so long it consumes the work

The proposition is where briefs live or die. Single-minded means one idea, and the test is mechanical: if you can remove the word and from the proposition and the sentence still says everything you meant, it was single-minded. If removing and loses something, you have written two propositions and delegated the choice to whoever makes the ad. That is not creative freedom. It is abdication, and it produces work that says two things weakly instead of one thing strongly.

Three further tests separate a good brief from a long one. Could this brief describe a competitor? If so it is not a brief, it is a category description. Does the objective have a number and a date? Without them nobody can say afterward whether the work succeeded, which suits everyone in the short run and nobody in the long run. And finally, does the brief inspire anybody? A brief is a persuasive document aimed at the creative team, and a technically complete brief that bores the people who have to act on it has failed at its actual job.

Key idea: A brief is single-minded when deleting the word and from the proposition costs nothing, and it is a real brief only when it could not equally describe a competitor.

Brand identity, image, and architecture

Three terms get confused constantly. Brand identity is what the organization intends the brand to mean and is entirely within its control. Brand image is what the audience actually believes, and is not. Brand equity is the commercial value created when image is strong, favorable, and consistent. The gap between identity and image is the honest measure of how a brand strategy is going, and closing it is most of the work.

Brand architecture is the question of how many brands an organization should have and how they relate. A house of brands, the Procter and Gamble model, keeps corporate identity in the background so each product brand stands alone, which contains risk and permits competing in the same category several times. A branded house, closer to the Google or Virgin model, puts one name on everything, which is efficient and transfers goodwill but also transfers damage. Most large companies run a hybrid, and the practical question is always whether the parent's endorsement helps or hurts in a given category.

Brand purpose, the idea that a brand should stand for a social mission beyond its product, has been the dominant fashion of the last decade and deserves a careful verdict. When a purpose reflects something the company actually does, it can organize decisions and attract staff. When it does not, audiences detect the gap quickly, and the reputational cost of claiming values you do not hold is higher than the cost of never claiming them. Most products do not need a purpose. They need to be good and to be remembered.

Key idea: Identity is what you intend, image is what people believe, equity is the value of the gap being small, and a claimed purpose that the business does not live costs more than no purpose at all.

Two schools of brand equity

Ask what actually builds a brand and you get two serious answers that disagree in useful ways.

The first is the associative tradition, most fully worked out by Kevin Lane Keller as customer-based brand equity. It describes a pyramid: first the brand must be salient, meaning people know who you are and in what category; then performance and imagery, what it does and what it means; then judgments and feelings, what people think and how they respond; and finally resonance, an active relationship and loyalty. In this view equity lives in a network of associations in memory, and the strategic job is to build associations that are strong, favorable, and unique.

The second is the empirical tradition of the Ehrenberg-Bass Institute, associated with Byron Sharp, which argues from large-scale purchase data that most of the loyalty story is overstated. Its claim is that brands grow through mental availability, the likelihood of being thought of in a buying situation, and physical availability, the ease of actually finding and buying the thing. On this account differentiation matters far less than practitioners believe, and what matters is distinctiveness: being instantly recognizable through consistent colors, logos, characters, sounds, and phrases, which this school calls distinctive brand assets.

The distinction between differentiation and distinctiveness is the single most useful idea in this section, so make it concrete. Differentiation is a reason to choose you. Distinctiveness is the ability to be identified as you in half a second, which is roughly all the attention any advertisement gets. A shopper who cannot tell your advertisement from a competitor's has given you nothing, no matter how good your reason to choose was. In practice the schools overlap more than their advocates admit: both agree that consistency over years is what compounds, and that changing your look because the marketing team is bored is a way of destroying value you already own.

Key idea: Keller's tradition builds equity through unique associations, while Ehrenberg-Bass builds it through mental and physical availability plus distinctive assets; both agree consistency over years is what compounds.

Common misconceptions

• Positioning is about listing everything good about the product. Positioning is subtraction: choosing one thing to be known for and giving up the rest.
• An empty space on a perceptual map is an opportunity. It may be empty because nobody wants what is there; the gap is a hypothesis to test, not a strategy.
• A brief should give the creative team every relevant fact. A brief that could describe a competitor has failed, and length is not thoroughness.
• Brand equity means loyalty. Purchase data shows most buyers are light and disloyal; availability in memory and in the store explain more of brand growth than devotion does.

Recap

• Positioning is what you do to the mind of the prospect, defined against competitors.
• Points of parity qualify you to compete; a point of difference is what the position is made of.
• Perceptual maps must be drawn from customer perception, not from the client's self-description.
• A brief is single-minded when deleting the word and from the proposition costs nothing.
• Identity is intended, image is believed, and equity is the value of a small gap between them.
• Distinctiveness, being recognizable instantly, often matters more than differentiation, being different in kind.

Sources

1. Wikipedia contributors. (n.d.). Positioning (marketing). Wikipedia. en.wikipedia.org
2. Wikipedia contributors. (n.d.). Brand equity. Wikipedia. en.wikipedia.org
3. Ehrenberg-Bass Institute for Marketing Science. (n.d.). How brands grow: mental and physical availability. marketingscience.info
4. Encyclopaedia Britannica. (n.d.). Brand. britannica.com
5. Institute of Practitioners in Advertising. (n.d.). Effectiveness resources. IPA. ipa.co.uk

The big picture

Ask most people how advertising persuades and you get one of two wrong answers. The first is that it presents reasons and people weigh them, which would make advertising a form of debate. The second is that it works by hidden manipulation on a helpless mind, which would make it a form of hypnosis. The research supports neither. Persuasion is a set of ordinary mental processes that operate in plain sight, that you can name once you know what to look for, and that work on you even when you can name them.

That last point matters and it is worth pausing on. Knowing that a shop uses a scarcity cue does not make the cue stop working, any more than knowing about an optical illusion makes the lines look equal. Awareness helps at the margin, mostly by giving you a moment to check, but the honest position is that these mechanisms are features of how attention and memory work rather than tricks played on the inattentive.

This lesson gives you the models that hold up. The elaboration likelihood model, which explains when careful argument works and when it does not. Framing, which shows that the same fact stated two ways produces different decisions. The social influence principles Robert Cialdini catalogued. Mere exposure and fluency, which explain why repetition is not merely lazy. And the long argument about emotion versus information, which has better evidence behind it than most people assume.

Key idea: Persuasion works through ordinary processes of attention, memory, and judgment, which is why naming a technique does not switch it off.

The elaboration likelihood model

Richard Petty and John Cacioppo's elaboration likelihood model, developed through the 1980s, is the most useful single framework in this field. Its claim is that persuasion travels one of two routes depending on how much mental effort the audience puts into the message. On the central route, people attend to the substance of the argument, think about it, and are persuaded by its quality. On the peripheral route, people do not engage with the argument and are influenced instead by cues attached to it: who is speaking, how attractive they are, how many arguments there are regardless of quality, how the message sounds, how it feels.

Which route you get depends on two conditions, and both must be met for central processing. The audience must be motivated, meaning the decision matters to them personally, and they must be able, meaning they have the knowledge, time, and quiet to think. Miss either condition and processing falls to the peripheral route by default. Now consider the actual circumstances in which advertising is received: on a phone, between two other tasks, about a product costing nine dollars, for two seconds. Peripheral processing is not an occasional case in advertising. It is the normal case.

The model's most consequential prediction concerns durability rather than strength. Attitudes formed on the central route are more persistent, more resistant to counterargument, and better at predicting behavior. Attitudes formed peripherally are real but shallow and decay quickly, which is why a brand built entirely on peripheral cues has to keep paying to maintain them. This gives a genuinely useful planning rule: for high-involvement, expensive, considered purchases such as a mortgage or a car, argument quality is worth investing in, because the audience will actually elaborate. For low-involvement everyday goods, argument quality is largely wasted, and being memorable, likeable, and instantly recognizable is doing the real work.

Key idea: Central-route persuasion requires both motivation and ability to think, and since advertising is usually received with neither, peripheral cues carry most of the load; but centrally formed attitudes last far longer.

Framing

Amos Tversky and Daniel Kahneman demonstrated in 1981 that the way a choice is described changes the choice people make, even when the described outcomes are logically identical. Their Asian disease problem is the standard demonstration: participants told a program would save two hundred of six hundred people chose it strongly, while participants told the same program meant four hundred would die largely rejected it. Same arithmetic, opposite decisions. The general finding is that people are risk averse when outcomes are framed as gains and risk seeking when the same outcomes are framed as losses.

Advertising uses framing constantly, and once you see it you cannot unsee it. Ninety five percent fat free rather than contains five percent fat. Save forty dollars rather than pay one hundred and sixty. Nine out of ten dentists recommend rather than one in ten does not. Loss framing is used where the goal is action about risk, as in insurance and security, while gain framing dominates categories about pleasure and self-improvement. Both are entirely truthful, which is exactly what makes framing an ethical grey zone rather than a legal one: no factual claim is false, and yet the impression created is engineered.

A related family of effects concerns reference points. A price looks high or low only against something, so an anchor is provided: a crossed-out original price, a premium version placed beside the one you are meant to buy, a competitor's figure. The decoy effect goes further, adding a deliberately unattractive third option to make a target option look reasonable. These are among the most reliable findings in behavioral economics, and they explain a great deal of what is happening on any pricing page you have ever seen.

Key idea: Logically identical facts produce different decisions depending on whether they are framed as gains or losses, which makes framing entirely truthful and still capable of engineering an impression.

Cialdini's principles of influence

Robert Cialdini's synthesis of the social influence literature identified six principles that recur across contexts, with a seventh added later. They are worth memorizing because they describe most of what persuasion practitioners actually do.

Principle	Mechanism	In advertising and PR
Reciprocity	We feel obliged to return favors	Free samples, free tools, useful content marketing
Commitment and consistency	We act consistently with past commitments	Free trials, small first steps, loyalty programs
Social proof	We look to others to decide what is correct	Reviews, follower counts, best seller labels
Authority	We defer to credible expertise	Expert endorsement, clinical language, certifications
Liking	We say yes to people we like	Likeable spokespeople, humor, similarity to the audience
Scarcity	We value what is limited	Limited editions, countdown timers, only three left
Unity	We favor those in our shared identity	Community language, in-group signals, fan identity

Social proof is the most heavily used and the most abused, because it can be faked more easily than the others. A genuine review is evidence; a purchased review is fraud, and the FTC has brought cases on exactly this point. Scarcity is next, and a countdown timer that resets when you reload the page is not a scarcity cue, it is a false statement of fact. The line here is the one from Lesson 1: using a real principle is persuasion, manufacturing the underlying fact is deception.

Authority deserves a specific note because it interacts with disclosure. A dermatologist recommending a moisturizer is authority working legitimately if the audience knows the relationship, and deception if a paid endorsement is presented as independent professional advice. This is precisely why the FTC endorsement guides require disclosure of material connections, a rule we will apply to hard cases in Module 6.

Key idea: Cialdini's principles describe most persuasion practice, and each becomes deceptive at the same point: when the underlying fact the principle relies on is manufactured or the connection concealed.

Repetition, exposure, and fluency

Repetition is the most mocked feature of advertising and one of the best evidenced. Robert Zajonc demonstrated in 1968 that simply encountering a stimulus repeatedly increases liking for it, with no argument attached and no memory of the exposures required. The mere exposure effect is robust, it works below conscious awareness of the exposures themselves, and it is a large part of why a brand pays to be seen rather than to say something.

The mechanism most researchers now favor is processing fluency. Things encountered before are easier for the brain to process, that ease feels good, and we misattribute the good feeling to the thing rather than to the ease. The same mechanism explains why simpler brand names, clearer typography, and rhyming slogans outperform their complicated equivalents: fluency is mistaken for truth, a phenomenon documented as the illusory truth effect, in which repeated statements are rated as more likely to be true regardless of their accuracy. That finding should worry you, because it applies as much to political claims and health misinformation as to slogans.

A related and useful model comes from Robert Heath's work on low attention processing, which argues that advertising does much of its work at very low levels of attention, building associations that are never consciously evaluated and therefore never counterargued. That explains an otherwise puzzling fact: advertising you actively ignore can still affect you, and advertising you actively dislike can still work, because dislike requires the attention that would also permit resistance.

Key idea: Mere exposure and processing fluency make repeated things feel better and truer, which is why advertising can work at attention levels too low for you to argue back.

Emotion, information, and fear

The oldest argument in the business is whether advertising should make you feel or make you think. There is real evidence, and it favors feeling more than most students expect. Analyses of the Institute of Practitioners in Advertising effectiveness databank by Les Binet and Peter Field found that campaigns using predominantly emotional approaches produced larger long-term business effects than predominantly rational ones, while rational, information-led messages performed better at driving immediate short-term response. That is not a contradiction; it is a division of labor, and it is the empirical backbone of the long and short framework we return to in Module 5.

The mechanism is not mysterious. Emotional content is better encoded and better remembered, it is processed adequately at low attention, and it attaches feelings to a brand that surface later at the moment of choice without any recalled argument. Informational content requires the elaboration that the elaboration likelihood model says is usually absent, but when the audience is genuinely in the market, motivated and able to think, information is exactly what they want and emotional work alone will frustrate them.

Fear appeals need separate and careful treatment, because they are the most misused tool in public communication. The meta-analytic evidence, notably Kim Witte and Mike Allen's review of the fear appeal literature, is that fear increases persuasion only under a specific condition: the audience must also receive an efficacy message telling them a specific action that works and that they can perform. Fear plus efficacy changes behavior. Fear without efficacy produces defensive avoidance, denial, and reactance, meaning people reject the message to restore their sense of control. Graphic anti-smoking or road safety campaigns that terrify without offering a doable step frequently fail for exactly this reason, and understanding it is the difference between a public health campaign that works and one that merely upsets people.

Key idea: Emotional approaches build larger long-term effects while informational ones drive short-term response, and fear only persuades when it is paired with a specific, achievable action.

Common misconceptions

• Advertising works by subliminal manipulation. The famous 1957 cinema experiment was later admitted to be fabricated, and the reliable effects are ordinary ones like exposure, fluency, and framing operating in plain sight.
• If you can name the technique, it stops working. Naming an effect gives you a moment to check, but exposure, fluency, and framing operate on people who can explain them perfectly.
• Better arguments always persuade better. Argument quality matters only when the audience is both motivated and able to elaborate, which is rare in advertising conditions.
• Scarier messages change behavior more. Fear without an efficacy message produces avoidance and reactance rather than action.

Recap

• The elaboration likelihood model splits persuasion into a central route driven by argument and a peripheral route driven by cues.
• Motivation and ability determine the route, and advertising conditions usually supply neither.
• Framing changes decisions without changing facts, and anchors and decoys shape how prices are read.
• Cialdini's principles describe practice; each becomes deception when the underlying fact is manufactured.
• Mere exposure and processing fluency make repeated things feel better and seem truer.
• Emotional work builds long-term effects, informational work drives short-term response, and fear needs efficacy to function.

Sources

1. Wikipedia contributors. (n.d.). Elaboration likelihood model. Wikipedia. en.wikipedia.org
2. Tversky, A., and Kahneman, D. (1981). The framing of decisions and the psychology of choice. Science, 211(4481), 453-458. doi.org
3. Zajonc, R. B. (1968). Attitudinal effects of mere exposure. Journal of Personality and Social Psychology, 9(2, Pt. 2), 1-27. doi.org
4. Witte, K., and Allen, M. (2000). A meta-analysis of fear appeals: Implications for effective public health campaigns. Health Education and Behavior, 27(5), 591-615. doi.org
5. Wikipedia contributors. (n.d.). Robert Cialdini. Wikipedia. en.wikipedia.org

The big picture

Here is a claim you will hear in a client meeting: our campaign drove a twelve percent sales lift. Here is the question almost nobody asks: compared with what? Compared with the same weeks last year, when the weather was different? Compared with regions that got no advertising, but also have different stores? Compared with people who did not see the ad, who differ from those who did in the single most important way, namely that the targeting system selected them?

That last problem is the heart of this lesson and it has a name. Targeting systems are designed to show advertisements to people most likely to buy. If you then compare buyers among the exposed with buyers among the unexposed, you are measuring the quality of the targeting, not the effect of the advertising. This is selection bias, it is enormous, and it contaminates a very large share of what the industry reports as effectiveness.

What follows is the honest version of what advertising does. The effects are real. They are also smaller than industry claims, slower than quarterly reporting cycles, and much more expensive to measure reliably than anyone wants to admit. I am going to give you numbers from peer-reviewed meta-analyses and large field experiments, and I want you to hold two things at once: advertising works, and most published measurements of how well it works are overstated.

Key idea: Because targeting selects the people most likely to buy anyway, comparing exposed with unexposed measures the targeting rather than the advertising, and that single confusion inflates most reported results.

How big is the effect

The standard summary measure is advertising elasticity: the percentage change in sales associated with a one percent change in advertising spend. Raj Sethuraman, Gerard Tellis, and Richard Briesch pooled hundreds of estimates in a 2011 meta-analysis and reported a mean short-term brand advertising elasticity of roughly 0.12, with a long-term figure around 0.24. They also found the short-term figure had declined over time compared with earlier decades.

Sit with that number, because students routinely misread it in both directions. An elasticity of 0.12 means that doubling advertising spend, a hundred percent increase, is associated on average with about a twelve percent increase in sales. That is not nothing; for a large brand it can be a great deal of money. It is also nowhere near the transformative effect implied by most campaign case studies, and it means that a modest budget increase produces a change often too small to detect against normal sales variation.

Two qualifications keep this honest in the other direction. First, averages hide enormous variance: new products, new categories, and genuinely distinctive creative work show much larger effects than established brands in mature categories. Second, elasticity measures the marginal effect of more spending on top of existing spending, which is a different question from what would happen if a brand stopped advertising entirely. The evidence on full withdrawal, including long-run studies of brands that went dark, suggests decline is slow at first and then accelerates, which is exactly the pattern that makes cutting the budget look free for a year.

Key idea: Meta-analysis puts mean short-term advertising elasticity near 0.12 and long-term near 0.24, meaning doubling spend typically moves sales by low double digits, far less than case studies imply.

Why measurement is so hard

Randall Lewis and Justin Rao made the statistical problem explicit in a paper whose title says it all: the unfavorable economics of measuring the returns to advertising. Their argument runs like this. Advertising effects are small relative to the variance in individual purchasing behavior. Detecting a small effect against a noisy background requires a very large sample. They showed, using a series of large-scale online field experiments, that even experiments with millions of subjects often cannot distinguish a campaign with a healthy return on investment from one that lost money.

Notice how uncomfortable that conclusion is. It does not say advertising fails. It says that for many realistic campaigns, the honest confidence interval around the return is wide enough to include both a good investment and a bad one, and no amount of dashboard sophistication closes that gap. Any measurement approach that reports a precise return without an experiment is producing precision it has not earned.

Two further problems compound this. Advertising effects decay over time in a pattern usually modeled as adstock, meaning a share of this week's exposure carries into following weeks, so effects measured in a short window are systematically understated. And selection is everywhere. People who see an advertisement chose the media, the site, the moment, and were chosen by an algorithm, which means observational comparisons are almost never comparing like with like.

Key idea: Lewis and Rao showed that advertising effects are small relative to purchase-behavior noise, so even experiments with millions of subjects often cannot separate a profitable campaign from a losing one.

What the experiments found

Field experiments are the only reliable answer, and the good ones have been unsettling. Three deserve to be known by name.

Leonard Lodish and colleagues analyzed 389 real-world split-cable television experiments in 1995, in which matched households in the same market received different advertising. Their headline finding was that increasing television advertising weight for established brands often produced no detectable sales effect, though new products responded more, and that when effects did occur they persisted for a considerable period afterward. That result, from the medium and era everyone assumed was well understood, should have caused more alarm than it did.

Thomas Blake, Chris Nosko, and Steven Tadelis ran a large experiment at eBay published in 2015, switching off paid search advertising in matched regions. For branded keywords, meaning searches containing the word eBay, the return was essentially zero: people searching for eBay went to eBay whether or not eBay paid for the top slot, because the free organic result was right below the ad. Effects for non-branded search were positive but far below the company's internal estimates. The observational methods in use had massively overstated returns, precisely because searchers who click a brand's ad were already looking for that brand.

Brett Gordon and colleagues, publishing in 2019, compared measurement approaches against fifteen large randomized experiments run on Facebook. Common observational methods, including matching and regression adjustments on rich individual-level data, frequently produced estimates far from the experimental benchmark, sometimes several times too large and sometimes in the wrong direction. Their conclusion was blunt: even with unusually detailed data, observational methods are not a reliable substitute for randomization.

Key idea: The eBay search experiment found near-zero incremental return on branded keywords, and the Facebook comparison found observational methods often wrong by multiples, so randomized holdouts are the only trustworthy measurement.

The spend-cut episodes, treated factually

Two widely discussed corporate episodes belong here, and they are worth stating carefully rather than dramatically, because both get exaggerated in both directions.

Procter and Gamble, among the largest advertisers in the world, publicly pressed the digital advertising industry on transparency, viewability, fraud, and agency contracts from January 2017 onward, and subsequently reported cutting well over one hundred million dollars of digital spend in a single quarter after concluding that a portion of it was reaching low-quality or ineffective inventory. The company reported that reach was not damaged, and it framed the exercise as removing waste rather than as evidence that digital advertising does not work. Both parts of that framing matter: the cut was real, and the company continued to spend heavily on digital afterward.

Uber's case is a fraud story with a measurement lesson attached. In 2017 Uber sued Fetch Media, its mobile advertising agency, alleging it had been charged for fraudulent and worthless app-install advertising, including installs credited to advertising that had never actually been shown. When Uber shut off a very large share of its app-install budget during the dispute, it reported that installs did not fall meaningfully, because much of the spending had been claiming credit for downloads that would have happened anyway.

Draw the right lesson from both. Neither episode shows that advertising is ineffective. They show that attribution which credits an advertisement whenever an exposed person later converts will keep reporting success even when the advertising is being served to bots, placed on worthless inventory, or shown to people who had already decided. The failure is in the measurement model, and that is exactly what Module 5 takes apart.

Key idea: The Procter and Gamble and Uber spend cuts revealed broken attribution and wasted inventory rather than proving advertising does not work, and in both cases spending continued after the waste was removed.

Short and long, and why the argument matters

The final piece is timing. Les Binet and Peter Field's analysis of the IPA effectiveness databank distinguishes short-term sales activation, which produces a sharp spike that decays within weeks, from long-term brand building, which accumulates slowly and raises the base level of demand over years. Their well-known working ratio, around sixty percent of budget to brand building and forty percent to activation for a typical business, is a rule of thumb rather than a law, and it varies by category, but the underlying structure is well supported.

Here is why this is not an academic quarrel. Short-term activation is easy to measure, because the effect arrives inside the reporting window and can be attributed to a click. Long-term brand building is hard to measure, because the effect arrives after the marketing director has been promoted or replaced. When you reward what is measurable, budgets drift toward activation, and the drift looks efficient every single quarter while slowly starving the thing that makes activation cheap in the first place. A brand with high mental availability converts activation spending far more efficiently than one without it.

So the honest summary of this module is a paradox you should be able to hold. Advertising works, in aggregate, over years, in ways well documented by experiments. Any specific claim about any specific campaign's return should be treated as unproven unless it came from a randomized holdout, and most do not. Learning to say both sentences without flinching is what separates a professional from an enthusiast.

Key idea: Short-term activation is easy to measure and easy to over-fund, while long-term brand building is hard to measure and is what makes activation efficient, so measurable does not mean valuable.

Common misconceptions

• Digital advertising is precisely measurable. It produces many numbers, but observational comparisons measure targeting quality rather than advertising effect.
• A sales lift after a campaign proves the campaign worked. Without a randomized control there is no counterfactual, and seasonality, price, distribution, and competitor activity all move sales.
• The eBay and Uber results show advertising does not work. They show attribution was crediting advertising for conversions that would have happened anyway.
• Small elasticities mean advertising is not worth doing. The average conceals large variance, and elasticity measures marginal spend rather than the effect of going dark entirely.

Recap

• Mean short-term advertising elasticity is around 0.12, with long-term around 0.24.
• Advertising effects are small relative to purchasing noise, so reliable measurement needs very large randomized samples.
• Split-cable television experiments often found no detectable effect for established brands.
• eBay's experiment found near-zero incremental return from branded paid search.
• Observational methods, even on rich data, frequently miss the experimental benchmark by multiples.
• Short-term activation is measurable and over-funded; long-term brand building is what makes activation efficient.

Sources

1. Sethuraman, R., Tellis, G. J., and Briesch, R. A. (2011). How well does advertising work? Generalizations from meta-analysis of brand advertising elasticities. Journal of Marketing Research, 48(3), 457-471. doi.org
2. Lewis, R. A., and Rao, J. M. (2015). The unfavorable economics of measuring the returns to advertising. Quarterly Journal of Economics, 130(4), 1941-1973. doi.org
3. Blake, T., Nosko, C., and Tadelis, S. (2015). Consumer heterogeneity and paid search effectiveness: A large-scale field experiment. Econometrica, 83(1), 155-174. doi.org
4. Gordon, B. R., Zettelmeyer, F., Bhargava, N., and Chapsky, D. (2019). A comparison of approaches to advertising measurement: Evidence from big field experiments at Facebook. Marketing Science, 38(2), 193-225. doi.org
5. Lodish, L. M., et al. (1995). How T.V. advertising works: A meta-analysis of 389 real world split cable T.V. advertising experiments. Journal of Marketing Research, 32(2), 125-139. doi.org

The big picture

A creative director once told a team that their work was not an idea, it was a joke about the product, and that the difference would show up in six weeks when they needed forty more of them. That is the most useful definition I know. An idea is a thought that keeps generating; an execution is a single finished thing. If you cannot make a second and a fifteenth and a fortieth from the same thought, without repeating yourself and without straining, you have a nice execution and no campaign.

The practical test is exactly that: can this thought produce many different pieces of work, in different formats, over years, all recognizably from the same place? Absolut's bottle-shape campaign passed that test for decades. A single funny film that cannot be extended does not, however good it is. Neither test is about quality. Both are about generativity, which is what makes an idea worth the money it takes to establish.

This lesson is about craft, and craft is the part of the field that most resembles a trade. There are rules, most of them learned by making things badly first. I will give you the ones that are actually load bearing: how concepting works as a process, what makes copy good, what art direction is doing, and how to read the industry's awards without being fooled by them.

Key idea: An idea is generative and an execution is finished; if a thought cannot produce forty different pieces of work, it is not the campaign idea.

Concepting as a process

Creative work is not waiting for lightning. It is a process with two phases that must be kept apart, and the most common failure in a creative session is running them at the same time.

The divergent phase generates quantity without judgment. The rule is volume: fifty rough thoughts, written badly, none defended, nothing dismissed. Judgment during generation kills the fragile early forms that good ideas often arrive in, and it makes the room politically unsafe in a way that reliably produces conventional work. Useful techniques for this phase include attacking the problem from deliberately different angles, the product, the user, the enemy, the occasion, the category convention you could break, and writing the headline you would be embarrassed to show anyone, because it often contains the real thought.

The convergent phase applies judgment ruthlessly, and it uses criteria set before the ideas existed so that they cannot be bent to protect a favorite. Three questions do most of the work. Is it on brief, meaning does it deliver the single proposition rather than something adjacent that happens to be more fun. Is it generative, by the forty-executions test. And is it ownable, meaning would it still work if a competitor's logo were dropped in, in which case it is a category idea and you are about to spend your budget advertising the whole market.

James Webb Young's old five-step account of the process still holds up: gather raw material, work it over mentally, drop it and do something else, wait for the idea to arrive, then develop and test it against reality. The unglamorous fourth step, deliberate incubation, is real and is supported by the psychology of insight. Ideas do arrive in showers, but only to people who did the first two steps properly.

Key idea: Keep generation and judgment strictly separate, then judge against criteria fixed in advance: on brief, generative, and ownable.

Copywriting

Copywriting is not writing prettily. It is writing to make something happen, under severe constraints of attention, with every word paying rent. Six principles carry most of the craft.

First, specificity beats adjectives. Delicious is a claim about your opinion; still warm when it reaches the table is a picture in the reader's head. The general rule is that concrete nouns and strong verbs do the work adjectives pretend to do. Second, write to one person. Advertising addressed to an audience sounds like an announcement; addressed to a person it sounds like speech, which is why second person and contractions outperform the corporate plural. Third, read it aloud. If you run out of breath or stumble, so will the reader, and radio and social audio are unforgiving of sentences built for the eye.

Fourth, cut. The first draft is for finding the thought and the second is for removing everything that was scaffolding. Most headlines improve when the first three words are deleted, because writers habitually clear their throat before saying the thing. Fifth, the headline and the visual should not repeat each other. If the picture shows a mountain and the headline says mountain, one of them is wasted; the good pairing makes the reader do a small piece of work to connect them, and that small work is what makes the message stick. Sixth, know what you want the reader to do, and say it. A call to action is not a formality, and a beautiful piece of communication that leaves the reader with nowhere to go has failed at the only thing that gets measured.

One structural note about long copy, since students assume nobody reads it. People read as much as they are interested in, and no more, which means length is a function of involvement rather than of format. Ogilvy's long-copy print advertisements worked because their readers were genuinely considering a purchase. The same length in front of an uninterested audience is not thoroughness; it is noise.

Key idea: Copy earns attention through specificity, direct address, and ruthless cutting, and headline and visual should complete each other rather than repeat each other.

Art direction

Art direction is the visual half of the same thought, and Bernbach's insight was that it has to be conceived alongside the words rather than applied to them afterward. Four ideas will let you read and judge visual work usefully.

Hierarchy: the eye needs somewhere to land first, then a path. A layout in which everything is emphasized has no emphasis, which is why the most common fault in amateur work is three competing focal points and a logo fighting all of them. Contrast: difference in size, weight, color, or space is what creates hierarchy, and space is the element beginners are most reluctant to use. The empty white page of the Volkswagen advertisements was not modesty. It was the loudest thing on a magazine spread full of chrome.

Typography carries tone before a single word is read, which is why a typeface choice is a strategic decision and not a decorative one. And branding, in the sense of Module 2's distinctive assets, is the discipline of making the work identifiable as yours within about half a second, using consistent color, character, shape, and sound rather than by making the logo bigger. The logo argument is almost always a symptom: a client asking for a larger logo is usually telling you the work is not recognizably theirs, and the fix is distinctiveness rather than size.

Finally, hold the practical constraint in mind while you design. Most work will be seen small, in a feed, at speed, often with sound off, cropped by a format you did not choose. Work that only functions at full size on a presentation screen is work that has not been designed for the conditions it will actually meet.

Key idea: Art direction creates hierarchy through contrast and space, carries tone through typography, and makes work identifiable in half a second through consistent distinctive assets rather than a bigger logo.

Structure and constraint

Two further craft points are worth naming because they explain why some work feels effortless and some feels laboured.

Narrative structure earns attention cheaply. A film with a setup, a complication, and a turn holds people because the brain wants resolution, which is why the six-second bumper is genuinely hard: it has room for one beat, so the entire idea must arrive at once or not at all. Structure is also what makes a demonstration work. Showing the product solving the problem, in that order, is the oldest reliable structure in the business and it still outperforms most cleverness.

Constraint is a creative asset, not an obstacle, and this is not a consoling platitude. A blank brief produces vague work because there is nothing to push against. The most productive constraints are a genuinely single proposition, a fixed format, a real budget limit, and a mandatory element. The Volkswagen work was constrained by an unfashionable small car and a small budget, and both constraints appear in the finished thought. When a creative team asks for more freedom, the useful response is usually a sharper brief rather than a looser one.

Key idea: Narrative structure buys attention cheaply, and tight constraints produce sharper work than open ones, which is why a vague brief is a creative problem and not a creative gift.

Awards, and what they measure

The industry runs on awards, and you need to read them accurately. Cannes Lions is the largest creative festival and its juries reward craft, originality, and idea quality as judged by peers. The Effie Awards judge effectiveness, requiring evidence of business results. Those are different questions, and confusing them is a standard beginner's error: a Cannes Lion is a judgment about the work and an Effie is a claim about what the work did.

The honest criticisms are worth knowing. Awards juries see work out of context, in a reel, without the media plan, which favors the striking single execution over the campaign that ran for four years. Entry fees mean well-funded agencies enter more. And the industry has a persistent problem with scam or ghost advertising, work created mainly to win awards, run minimally or not at all for a real client, which festivals have repeatedly tried to police with mixed success.

The defensible position is that awards are a proxy for craft quality and nothing more, and that as a proxy they have some value: the IPA effectiveness analyses have found a positive relationship between creatively awarded campaigns and business results, plausibly because distinctive work gets noticed and remembered. That is a real finding and it is not a licence to treat a trophy as evidence about a specific campaign. Judge work by what it was asked to do.

Key idea: Cannes judges the work and Effie judges the results, and creative awards are a proxy for craft quality rather than evidence about a particular campaign's business effect.

Common misconceptions

• A great advertisement is an idea. An execution is a finished thing; an idea is what can generate forty more of them across formats and years.
• Creativity is inspiration rather than process. The reliable method separates divergent generation from convergent judgment, with incubation in between.
• Nobody reads long copy. People read as much as they are interested in; length should track involvement, not format convention.
• Making the logo bigger fixes weak branding. A demand for a bigger logo usually signals that the work lacks distinctive assets, which is a design problem, not a scale problem.

Recap

• The test of an idea is generativity: can it produce many executions across formats and years.
• Divergent generation and convergent judgment must be separated, and judgment must use criteria set in advance.
• Copy works through specificity, direct address, reading aloud, and cutting.
• Headline and visual should complete each other rather than repeat each other.
• Art direction builds hierarchy with contrast and space, and identity with consistent distinctive assets.
• Cannes judges craft and Effie judges effectiveness; awards are a proxy for quality, not evidence of results.

Sources

1. Encyclopaedia Britannica. (n.d.). Advertising. britannica.com
2. Cannes Lions International Festival of Creativity. (n.d.). Awards and judging criteria. canneslions.com
3. Effie Worldwide. (n.d.). About the Effie Awards. effie.org
4. American Advertising Federation. (n.d.). About the AAF and the American Advertising Awards. aaf.org
5. Wikipedia contributors. (n.d.). Think Small. Wikipedia. en.wikipedia.org

The big picture

An idea that exists only as a thirty-second film is not a campaign, it is a film. The idea has to survive being cut to six seconds, having its sound switched off, being cropped to a vertical rectangle, being read aloud with no picture at all, being printed at the size of a postage stamp, and being wrapped around a package in a shop. Formats are not delivery pipes into which a finished thing is poured. Each one changes what the work can be, and planning for that at the concept stage is the difference between a campaign and a set of increasingly sad adaptations.

The second half of this lesson is about testing, and I want to be direct: most creative testing in this industry is weak, and it is weak in a specific and predictable direction. It systematically favors work that is familiar, comfortable, and average, and it penalizes work that is unusual, which is the only kind that gets noticed. Understanding exactly how the machinery produces that bias is more useful than either defending testing or dismissing it.

By the end of this lesson you should be able to take one idea and plan its life across a real media mix, describe how it gets made and cleared, and read a testing report knowing which numbers mean something and which are artifacts of the method.

Key idea: Formats change what an idea can be, so a campaign has to be conceived for its whole media life, not adapted into it afterward.

One idea, many formats

Every format imposes three conditions: how long you have, how much attention you get, and what senses are available. Plan against those three and the adaptation problem becomes tractable.

Format	Typical attention	What it is good for	Main constraint
Long-form film	Chosen, sustained	Story, emotion, demonstration	Only reaches people who opt in
Thirty-second video	Interrupted, partial	Setup, complication, turn	Skippable after a few seconds
Six-second bumper	Fragmentary	One beat, one asset, one thought	No room for structure
Social feed video	Very low, often silent	Instant recognition, captions, motion	Vertical crop and sound off
Audio and podcast	Moderate, hands busy	Voice, humor, host read	No image at all
Out of home	Seconds, in passing	Big simple thought, one image	Word count and viewing distance
Print and long copy	High if involved	Detail, argument, reason to believe	Only works for interested readers
Packaging and retail	At the moment of choice	Recognition and reassurance	Fixed area, regulatory text

Two practical rules follow. The first is design for sound off but reward sound on. A large share of feed video is watched silently, so the idea must be legible without audio through captions, visible action, and on-screen type, while still rewarding the person who turns the sound up. The second is brand early and brand persistently. If your distinctive asset appears only at the end, the majority of viewers who leave before the end have watched an advertisement for the category.

Adaptation is not the same as translation. Cutting the thirty-second film down to six seconds usually produces an incoherent fragment, because the six-second form needs its own single beat rather than the compressed remains of a three-beat structure. The professional habit is to concept each key format separately from the same idea, then check they share assets and tone. For international work, the same logic applies at a larger scale: transcreation rewrites an idea to work in another language and culture, while literal translation reliably produces jokes that are not funny and idioms that mean nothing.

Key idea: Design for sound off, brand early, and concept each major format from the idea rather than compressing a finished film into smaller boxes.

How the work actually gets made

Between an approved concept and a live advertisement sits a production process that students rarely see and that consumes most of the budget and nearly all of the calendar.

The sequence runs roughly like this. Concepts are presented and one is approved, usually with changes. Pre-production covers casting, location, director selection through a bidding process among production companies, and a pre-production meeting where every detail is agreed in advance because changing your mind on a shoot day is extraordinarily expensive. The shoot happens. Post-production covers editing, color grading, sound design, music licensing, and visual effects. Then come the approvals: client sign-off, legal and regulatory clearance, and in some categories a network or platform review. Finally the asset is versioned for every format and trafficked, meaning delivered with the correct specifications to each media owner or platform.

Legal clearance deserves emphasis because it is where confident work goes to die and where careers occasionally end. Every factual claim must be substantiated before it runs, which means holding the evidence in a file. Comparative claims about competitors invite challenge at the National Advertising Division or in court. Music, images, fonts, and stock footage need licences for the specific territory, media, and duration, and a licence that covered a social post does not cover a television run. People appearing in the work sign releases. Health, financial, alcohol, and children's categories carry additional rules layered on top. None of this is optional, and the professional habit is to involve legal at the concept stage rather than the week before launch, because a claim that cannot be substantiated is cheaper to abandon as a sentence than as a finished film.

Key idea: Production runs from approved concept through pre-production, shoot, post, clearance, versioning, and trafficking, and legal substantiation belongs at the concept stage rather than the week before launch.

How creative testing is done

Copy testing, or pre-testing, means evaluating work before it runs. The main methods fall into four groups.

Qualitative exposure puts rough work in front of small groups and asks what they think. Recall and recognition tests expose respondents to work, often within a program or a mock feed, and measure the next day whether they remember the advertisement and the brand; day-after recall has been an industry standard for decades. Implicit and biometric measures try to get past what people say, using response-time tasks, facial coding, eye tracking, and similar tools to infer attention and emotion. And in-market experiments, including platform brand lift studies and randomized holdouts, measure real behavior among people who did or did not receive the advertising.

Those four differ enormously in what they can support. The first three are pre-tests conducted under artificial conditions; only the fourth measures anything about the real world. That distinction matters more than any argument between vendors, and it is the frame for everything that follows.

Key idea: Pre-testing methods measure reactions under artificial exposure, while only in-market experiments observe behavior under real conditions.

Why most testing is weak

Six problems recur, and they compound.

Forced exposure is the first and largest. A respondent sitting in a test environment, knowing they will be asked questions, watches an advertisement with attention nobody gives it in life. Lesson 6 explained why this matters: forced attention shifts processing toward the central route, so the test measures how the work performs under conditions of deliberate evaluation, which is precisely the condition the work will never meet.

Second, recall is not persuasion. An advertisement can be remembered vividly and sell nothing, and the perennial finding that people recall an amusing film while misattributing the brand is the clearest case. Third, articulation bias favors work that can be explained. Asked why they liked something, respondents produce reasons, and work whose effect is emotional or associative loses to work with a statable argument, even where the evidence in Lesson 7 says the emotional work will produce larger long-term effects.

Fourth, normative databases pull toward the average. Most commercial pre-testing scores work against a database of previously tested advertisements, which encodes what has been done before as the standard. Unusual work reliably scores below norm on measures like clarity and likeability at first exposure, and the honest question is whether the low score is a warning or exactly what novelty looks like. The Volkswagen work of 1959, small and plain against a category of chrome, is the standard reference for what such a system would probably have discouraged. Fifth, single exposure misrepresents repetition, since real campaigns work through accumulation and wear-in, and a first-exposure test cannot see the effect that only appears on the fifth encounter.

Sixth, and most damaging institutionally, testing is used as political cover. A number, however weakly grounded, lets a decision be attributed to research rather than to a person, which makes it very hard to overrule and very attractive to everyone who does not want to be responsible. That is not a flaw in the method; it is a flaw in how organizations use methods.

Key idea: Forced exposure, recall proxies, articulation bias, normative databases, and single exposure combine to make pre-testing favor familiar, average work, and organizations then use the score to avoid taking responsibility for a judgment.

What good testing looks like

None of this means testing is worthless. It means being clear about what a given test can support, and the useful distinction is diagnostic versus predictive.

Diagnostic testing asks what is happening in the work: is the brand identified, is the story followed, is anything confusing or offensive, does the joke land in this market, does the claim read as credible. Qualitative work and eye tracking are genuinely good at this, and the answers improve the work. Predictive testing asks whether this advertisement will produce more sales than that one. That is a causal question about the real world, and almost no pre-test can answer it, however confident the score looks.

So use pre-testing to fix problems, not to pick winners, and reserve the pick-winners question for in-market experiments: matched-market tests, randomized holdouts, and multi-cell designs that run several versions simultaneously against a control. And apply two habits. Always ask what the test's control group was, because a test without a control is a description rather than an experiment. And decide the decision rule before you see the numbers, in writing, because a threshold agreed afterward is not a threshold at all.

Key idea: Use pre-testing diagnostically to fix problems, and answer the which-works-better question with in-market experiments that have real control groups and a decision rule set in advance.

Common misconceptions

• A thirty-second film can simply be cut down to six seconds. The short form needs its own single beat; compressing a three-beat structure produces an incoherent fragment.
• Translating a campaign is enough for another market. Transcreation rewrites the idea to work in another language and culture; literal translation produces jokes that are not funny.
• A low pre-test score means the work is bad. Unusual work reliably scores below normative databases at first exposure, which is often what novelty looks like rather than a warning.
• High recall proves an advertisement works. Work is frequently remembered while the brand is misattributed, and recall is not persuasion.

Recap

• Every format sets its own conditions of duration, attention, and available senses.
• Design for sound off, brand early, and concept key formats separately from the same idea.
• Production runs concept, pre-production, shoot, post, clearance, versioning, trafficking, and legal belongs at the start.
• Pre-testing methods measure artificial exposure; only in-market experiments observe real behavior.
• Forced exposure, recall proxies, articulation bias, and normative databases bias testing toward average work.
• Test diagnostically to improve work, and use controlled in-market experiments to choose between options.

Sources

1. Interactive Advertising Bureau. (n.d.). Ad formats, specifications, and standards. IAB. iab.com
2. Federal Trade Commission. (n.d.). Advertising and marketing basics. ftc.gov
3. Gordon, B. R., Zettelmeyer, F., Bhargava, N., and Chapsky, D. (2019). A comparison of approaches to advertising measurement: Evidence from big field experiments at Facebook. Marketing Science, 38(2), 193-225. doi.org
4. Institute of Practitioners in Advertising. (n.d.). Effectiveness and testing resources. IPA. ipa.co.uk
5. BBB National Programs. (n.d.). National Advertising Division case decisions. bbbprograms.org

The big picture

A client has one million dollars. They can show an advertisement to one hundred thousand people ten times each, or to one million people once each, or to five hundred thousand people twice each. Every combination costs the same. Which is right? That single question is media planning, and everything else in this lesson is machinery for answering it.

Media planning decides where, when, and how often. Media buying then negotiates and executes it. The two used to be one job and in most agencies they are now different departments with different skills, because planning is a strategic argument and buying is a market operation. Both are far more consequential than students expect. The evidence in Lesson 7 suggests media decisions, particularly how widely you reach and how consistently you show up, account for a great deal of the variance in whether advertising works at all.

I want to warn you about one habit before we start. Media is the part of this field most heavily populated with numbers, and numbers create false confidence. Every metric in this lesson is a proxy for something we actually care about and cannot see directly. Keep asking what each number is standing in for, and you will avoid the most common professional error, which is optimizing a proxy until it stops representing the thing it was a proxy for.

Key idea: Media planning is the allocation of a fixed budget between reaching more people and reaching the same people more often, and that trade governs almost everything else.

The vocabulary

Six terms carry most of the conversation, and precision matters because they are routinely misused.

Term	Definition	What it hides
Impressions	Total exposures delivered, counting repeats	Says nothing about how many people
Reach	Unique people exposed at least once, often as a percent	Says nothing about how often
Frequency	Average exposures per reached person	An average across a very skewed distribution
Gross rating points	Reach multiplied by frequency	Two very different plans can share one number
Cost per thousand	Cost to deliver a thousand impressions	Cheap impressions are often cheap for a reason
Share of voice	Your advertising as a share of the category's	Depends on estimating competitor spend

The frequency average deserves the most suspicion. A campaign reporting an average frequency of four rarely means most people saw it four times. It usually means a small group of heavy media users saw it fifteen times while a much larger group saw it once, and the two ends of that distribution have completely different value. Always ask for the frequency distribution rather than the mean; if the plan cannot produce one, that itself is informative.

Cost per thousand invites a parallel error. Buying the cheapest available impressions is easy and almost always wrong, because price in an auction reflects what other buyers know about the value of that inventory. Very cheap impressions typically come from low-attention placements, unviewable positions, fraudulent traffic, or made-for-advertising sites built to arbitrage exactly this instinct. Efficiency measured on cost per thousand alone rewards buying worse media.

Key idea: Reach counts people, frequency counts repeats, and their average conceals a skewed distribution, while cheap cost per thousand usually signals inventory that other buyers correctly declined.

How much frequency, and when

The oldest argument in media is how many exposures are needed before an advertisement does anything. Herbert Krugman argued in 1972 for three, on a psychological account: the first exposure asks what is this, the second asks what of it, the third is a reminder. Effective frequency became doctrine, and plans were built to deliver three or more exposures to a target within a purchase cycle, deliberately sacrificing reach to do it.

Two later arguments largely dismantled that doctrine for established brands. Erwin Ephron's recency theory holds that what matters is being present close to the moment of purchase, since most categories have buyers entering the market continuously; on that account one exposure at the right time beats three at the wrong time, and continuous low-weight presence beats concentrated bursts. Separately, the Ehrenberg-Bass evidence in Lesson 4 argues that growth comes from reaching many light buyers, which means additional frequency to people already reached is usually worth less than the first exposure to someone new.

The current mainstream position, which is defensible rather than fashionable, is that reach beats frequency for established brands in mature categories, with continuous presence preferred to heavy flights. Frequency still earns its place in specific cases: new products that must teach something unfamiliar, complex propositions, low-attention environments where a single exposure is likely missed, and short campaigns with a fixed deadline such as an election or a film opening.

Scheduling patterns follow from this. Continuity spreads spend evenly all year and suits categories bought all year. Flighting concentrates spend into bursts with dark periods between, which is right when demand is genuinely seasonal. Pulsing keeps a low continuous base with bursts on top, and it is the common compromise, giving recency coverage without abandoning the ability to make noise when it matters.

Key idea: Krugman's three-exposure rule has largely given way to recency and reach arguments, so continuous broad presence usually beats concentrated frequency except for new or complex propositions.

How much to spend

Budget setting in practice is often embarrassing: a percentage of last year's sales, a match of what a competitor spends, or whatever is left after other costs. Each of these is circular or arbitrary, and the percentage-of-sales method is actively perverse, since it cuts advertising exactly when sales fall.

A better framework uses share of voice against share of market. The observed regularity, developed in the IPA effectiveness work by Binet and Field and consistent with earlier Nielsen analyses, is that brands whose share of voice exceeds their share of market tend to grow, and brands whose share of voice sits below their share of market tend to decline. The gap is called excess share of voice, and it converts an abstract budget question into a competitive one: what share of category advertising do we need to hold or to grow.

Two honest caveats. The relationship is an average across many cases with wide variation, so it sets a sensible expectation rather than a guarantee for any single brand. And it requires estimating competitor spend, which is imprecise. Even so, it is a considerable improvement on setting a budget by what is left over, and it makes explicit the fact that advertising is a competitive activity in which your effect depends on what everyone else is doing.

The related concept is diminishing returns. Response to advertising weight is not linear: reach curves rise steeply at first and flatten, so the tenth million dollars buys far less incremental reach than the first. A serious plan identifies roughly where the curve flattens for its category and stops there rather than pushing on because budget exists.

Key idea: Excess share of voice, spending a larger share of category advertising than your share of market, is associated with growth, and reach curves flatten so extra weight buys progressively less.

Choosing channels

Channel selection should follow from the job, and three questions settle most of it. What kind of attention does this message need, what does the audience actually use, and what does each channel cost per unit of the outcome you care about rather than per impression.

Broadly, high-attention, emotionally encoded, broadly reaching channels such as television, connected television, cinema, and audio suit brand building, while high-intent, addressable channels such as search, retail media, and direct response social suit activation. Out of home buys unavoidable public presence and credibility at scale, which is why brands use it when they want to look established. Print retains involved readers in specific categories. The point is not a hierarchy; it is matching the channel's attention conditions to the job the message is doing, which links directly back to the elaboration likelihood model in Lesson 6.

Paid media should also be planned alongside owned and earned rather than separately, because they compound. Paid drives people to owned channels where a fuller story can be told and a relationship started. Earned coverage lends credibility that paid media can then amplify by putting the third party's words in bought space. Owned assets like an email list reduce the cost of the next campaign because you no longer have to rent access to those people. Planning only the paid column is the single most common structural weakness in real media plans.

Key idea: Match a channel's attention conditions to the job the message is doing, and plan paid, owned, and earned together because each lowers the cost of the others.

Buying, and the numbers everyone argues about

Buying differs by medium. Broadcast television in the United States still runs a substantial upfront market, where inventory for the coming season is committed in advance at negotiated rates, with a scatter market for what is left. Out of home and print are negotiated by site, position, and duration. Digital is overwhelmingly bought through automated auctions, which the next lesson covers in detail.

Underneath all of it sits audience measurement, and this is where the arguments are fiercest. Panel-based measurement, historically Nielsen for television, extrapolates from a recruited sample and inherits all the strengths and weaknesses of sampling. Census-based digital measurement counts every event but counts only events it can see, which excludes devices, environments, and people it cannot track, and includes fraudulent events it cannot distinguish. Neither is neutral, and the difference between two vendors' numbers for the same campaign is often larger than the difference the campaign made.

Two further buying-side judgments matter. Brand safety and suitability decide what environments a brand refuses to appear beside, a set of choices with real editorial consequences: blunt keyword blocklists have repeatedly defunded legitimate news coverage of difficult subjects, which is a genuine harm to the information ecosystem rather than a technicality. And attention metrics, an active research area, attempt to measure how long an advertisement was actually looked at rather than merely served, on the reasonable view that viewability is far too low a bar.

Key idea: Panel measurement extrapolates from a sample and census measurement counts only what it can see, so vendor differences often exceed campaign effects, and blunt brand safety blocklists impose real costs on news publishers.

Common misconceptions

• An average frequency of four means people saw it four times. It usually means a few heavy users saw it many times and most people saw it once.
• The cheapest cost per thousand is the most efficient buy. Auction prices carry information, and very cheap inventory is usually low attention, unviewable, or fraudulent.
• Three exposures are needed before advertising works. Krugman's rule has largely given way to recency and reach arguments for established brands.
• Media planning means planning paid channels. Paid, owned, and earned compound, and planning only the paid column wastes the cheapest reach available.

Recap

• Reach counts unique people, frequency counts repeats, and gross rating points multiply the two.
• Average frequency conceals a highly skewed distribution, so ask for the distribution.
• Recency and reach arguments have largely displaced effective frequency for established brands.
• Excess share of voice is associated with growth, and reach curves show diminishing returns.
• Match channel attention conditions to the job, and plan paid, owned, and earned together.
• Panel and census measurement fail in different directions, and brand safety blocklists have real editorial costs.

Sources

1. Interactive Advertising Bureau. (n.d.). Measurement guidelines and standards. IAB. iab.com
2. Nielsen. (n.d.). Audience measurement. nielsen.com
3. Institute of Practitioners in Advertising. (n.d.). Share of voice and effectiveness resources. IPA. ipa.co.uk
4. Ehrenberg-Bass Institute for Marketing Science. (n.d.). Reach, frequency, and brand growth. marketingscience.info
5. Association of National Advertisers. (n.d.). Media transparency and measurement initiatives. ANA. ana.net

The big picture

You open a news site. Before the page finishes drawing, an auction has been held for the rectangle beside the headline, dozens of potential buyers have been asked whether they want to bid on showing something to you specifically, a winner has been chosen, payment terms have been settled, and the creative has been delivered. The whole thing took roughly a tenth of a second and nobody human was involved at any point. That is programmatic advertising, and about eighty to ninety percent of digital display in mature markets is now bought this way.

This lesson opens that black box. Not because you will personally configure a demand-side platform, but because almost every controversy in modern advertising, fees, fraud, privacy, brand safety, measurement, is a consequence of specific design decisions inside this machinery. People who cannot describe the auction cannot evaluate claims about it, and the industry contains a great many claims.

We will go through the programmatic stack, then search, then social platforms, then influencer marketing, which is the least automated and in some ways the least measurable channel in modern practice. Throughout, keep asking the question from the last lesson: what is each number a proxy for.

Key idea: Most digital display is now sold impression by impression in automated auctions, and the industry's disputes over fees, fraud, privacy, and measurement all follow from how that machinery is built.

The programmatic stack

Five components do the work, and knowing which is which makes the fee argument legible.

Component	Whose side	What it does
Demand-side platform	Advertiser	Decides what to bid on and how much, across many exchanges
Supply-side platform	Publisher	Offers inventory to buyers and manages floor prices
Ad exchange	Marketplace	Runs the auction matching bids to inventory
Ad server	Both	Delivers the creative and records what was served
Data platform	Advertiser	Supplies audience segments used to value an impression

The sequence is quick to state. A user loads a page. The publisher's supply-side platform sends a bid request describing the impression: the site, the position, the format, and whatever is known or inferred about the user and their device. That request reaches many demand-side platforms. Each decides in milliseconds whether this impression matches a campaign's targeting and what it is worth. Bids return, the exchange picks a winner, the winning creative is delivered from an ad server, and the page finishes loading.

Two design details matter more than they sound. Auctions were historically second price, meaning the winner paid slightly more than the second-highest bid, which encourages honest bidding; the market largely shifted to first price around 2019, meaning the winner pays what it bid, which changed bidding strategy considerably and made the arithmetic more transparent. And header bidding, introduced by publishers in the mid 2010s, lets a publisher solicit bids from multiple exchanges simultaneously before calling its primary ad server, which recovered some pricing power that the previous sequential waterfall had taken away.

Not everything runs in the open exchange. Private marketplaces restrict bidding to invited buyers at agreed floors, and programmatic guaranteed reserves specific inventory at a fixed price with automated delivery. Advertisers who care about environment and transparency have increasingly moved spend into these arrangements and toward curated inventory lists, precisely because the open exchange is where the fraud and made-for-advertising problems concentrate.

Key idea: A bid request describing the impression reaches many demand-side platforms, which bid in milliseconds; header bidding and the shift from second-price to first-price auctions were the two structural changes that most altered the market.

Search advertising

Search is different in kind because it sells intent rather than audience. A person typing a query has stated a need in their own words, which is why search advertising converts at rates no display placement approaches, and why it dominates digital spending.

The mechanism is an auction that is not decided on price alone. An advertiser bids on keywords, and position is determined by a combination of bid and an expected-quality assessment, which in Google's system is expressed as Quality Score and folded into ad rank. That design has a purpose worth understanding: a search engine that sold top position purely to the highest bidder would fill its results with irrelevant advertising and lose users, so relevance is priced in. The practical consequence is that better, more relevant advertising costs less per click than worse advertising at the same position.

The distinction that matters most strategically is branded versus non-branded keywords. Branded keywords contain your own name, and Lesson 7 gave the crucial evidence: eBay's field experiment found essentially no incremental return from bidding on them, because people searching for a brand find it anyway through the organic result immediately below. Non-branded keywords, where the searcher has described a need without naming a supplier, are where incremental value lives. Branded bidding is not always wrong, since it can be defensive when competitors bid on your name, but it should be justified rather than assumed, and it should be tested with a holdout.

One further caution. Search captures existing demand; it rarely creates it. A brand nobody has heard of will not be searched for, which is why search performance is downstream of the brand building in Module 3. Organizations that shift all their budget into search because it measures well eventually find their search volumes falling, and the cause is not the search program.

Key idea: Search sells intent and prices relevance into the auction, but it captures existing demand rather than creating it, and branded keyword bidding often has near-zero incremental value.

Social platform advertising

Social platforms sell an auction too, but with a different logic. You do not usually buy a placement; you state an objective, such as reach, video views, installs, or purchases, and the platform's optimization system decides who to show it to in order to maximize that objective. The targeting you specify is a constraint on the system rather than the mechanism itself, and the system will often outperform your targeting instructions if you give it room.

This produces a genuine strategic consequence that surprises people. Because the optimization is largely out of your hands, creative becomes the main controllable lever. Platforms and independent analyses consistently find that variation in creative explains far more variance in outcomes than the targeting parameters buyers spend most of their time adjusting. The practical implication is to invest in more and more varied creative rather than in ever finer audience definitions.

Two further points. Lookalike or similar audiences build a target from people resembling an existing customer list, which is powerful and also inherits every bias in the source list, including the tendency to keep finding customers who look like the customers you already have. And optimization toward an easily achieved objective will find easily achieved conversions: telling a system to maximize installs will find people who would have installed anyway, which is exactly the failure the Uber episode illustrated.

Two structural shifts since 2021 have reshaped this whole area. Apple's App Tracking Transparency required apps to ask permission before tracking users across other companies' apps and sites, and most users declined, which degraded cross-app measurement significantly. The long, repeatedly delayed retreat from third-party cookies in web browsers points the same way. The result is a measurable move toward first-party data, contextual targeting, and modelled rather than observed attribution, and it strengthens the case for the experimental methods in the next lesson.

Key idea: On social platforms you buy an objective rather than a placement, the optimization system rather than your targeting does most of the work, and creative variety is the main lever left to the buyer.

Retail media and connected television

Two fast-growing channels deserve naming because they are where budget is currently moving.

Retail media networks are advertising businesses run by retailers, selling sponsored placements on their own sites and apps, and increasingly off-site, using their purchase data. Their appeal is obvious: the retailer knows what people actually bought, which is the closest thing to ground truth in this whole field, and the advertising sits at the point of purchase. The caution is equally obvious, and it repeats a theme. The retailer both sells the advertising and reports on its effectiveness, using data no one else can audit, and a shopper who was already in your aisle is not a shopper you necessarily needed to pay to reach.

Connected television means streaming delivered over the internet, and it has inherited television budgets while behaving like digital. It offers the attention conditions of television with addressability, which is genuinely valuable, but it also imports the digital problems: fragmented measurement across platforms, frequency that is very hard to control across services that do not talk to each other, and the recurring result that a viewer sees the same advertisement far more often than any plan intended.

Key idea: Retail media offers purchase data at the point of sale but is measured by the party selling it, and connected television brings television attention with digital fragmentation and uncontrolled frequency.

Influencer marketing

Influencer marketing pays a person with an audience to feature a product. It works, when it works, for a reason Lesson 6 already gave you: it combines liking, authority, and social proof, and it arrives through a source the audience has chosen and feels they know. That perceived relationship, which researchers call parasocial, is doing the persuasive work, and it is exactly what is spent when it is misused.

Practitioners sort creators by audience size, from mega and macro influencers down through micro to nano creators with a few thousand followers. The consistent finding is that engagement rates fall as audience grows, so smaller creators often deliver more response per follower, though at higher management cost per campaign. Selection should turn on audience fit and authenticity rather than raw follower count.

Three problems have to be stated plainly. Fraud is widespread: followers, likes, and comments can be purchased cheaply, and a meaningful share of accounts show engagement patterns inconsistent with real audiences, so verification is a basic professional duty rather than an optional check. Measurement is weak, because most influencer reporting counts impressions and engagements, which are outputs, rather than incremental sales, and the same attribution problems from Lesson 7 apply with extra force. And disclosure is a legal obligation, not a courtesy.

On that last point the law is clear and worth memorizing now, since Module 6 applies it. The FTC endorsement guides require that any material connection between an endorser and a brand, which includes payment, free products, discounts, family relationships, and employment, be disclosed clearly and conspicuously in a way the audience will actually notice. Burying it in a hashtag block, placing it below a more link, or relying on a platform's small paid-partnership label alone has all been treated as inadequate. The obligation runs to the advertiser as well as the creator, which means an agency that fails to instruct and monitor its creators has a compliance problem of its own.

Key idea: Influencer marketing works through a parasocial relationship the audience chose, and it demands verification against follower fraud plus clear and conspicuous disclosure of any material connection, an obligation the advertiser shares.

Common misconceptions

• Programmatic buying means buying cheap remnant inventory. Programmatic is a transaction method, and premium inventory is routinely sold through private marketplaces and programmatic guaranteed deals.
• Bidding on your own brand name is obviously worthwhile. eBay's experiment found near-zero incremental return, since the organic result sits directly below the paid one.
• Better targeting is the main lever on social platforms. The optimization system does most of that work; creative variety explains more of the outcome variance.
• A paid-partnership label is always sufficient disclosure. The FTC requires clear and conspicuous disclosure that the audience will actually notice, and buried or platform-only labels have been found inadequate.

Recap

• A bid request describing an impression reaches many demand-side platforms, which bid within milliseconds.
• Header bidding and the shift from second-price to first-price auctions were the two big structural changes.
• Search sells intent and prices relevance into rank, but captures demand rather than creating it.
• Social buying means stating an objective and letting the optimization system find the audience, so creative is the main lever.
• App Tracking Transparency and cookie deprecation degraded observed attribution and pushed the field toward experiments.
• Influencer work requires fraud verification and clear, conspicuous disclosure of material connections.

Sources

1. Interactive Advertising Bureau. (n.d.). Programmatic and real-time bidding standards. IAB. iab.com
2. Wikipedia contributors. (n.d.). Real-time bidding. Wikipedia. en.wikipedia.org
3. Federal Trade Commission. (n.d.). Disclosures 101 for social media influencers. ftc.gov
4. Blake, T., Nosko, C., and Tadelis, S. (2015). Consumer heterogeneity and paid search effectiveness: A large-scale field experiment. Econometrica, 83(1), 155-174. doi.org
5. Pew Research Center. (n.d.). Social media fact sheet. pewresearch.org

The big picture

A person sees a television advertisement in March. In April a friend mentions the brand. In June they see three social posts. In July they search the brand name, click the paid result at the top, and buy. Question: what caused the sale? Under the attribution model that has governed digital marketing for two decades, the answer is the paid search click in July, which receives one hundred percent of the credit. The television, the friend, and the social posts receive nothing.

That is the last-click fallacy, and once you have seen it you will notice it everywhere. It is not merely imprecise; it is systematically biased in a particular direction. It over-credits the channels that appear closest to the purchase, which are usually the channels that harvest demand rather than create it, and it under-credits everything that built the demand in the first place. Organizations that optimize against it reliably shift budget from creating demand to harvesting it, watch their efficiency metrics improve, and slowly run out of demand to harvest.

This lesson is about how to know whether advertising worked. It is the hardest question in the field and it has no fully satisfying answer, but it has a set of honest ones, and knowing their limits is what separates useful analysis from expensive theatre.

Key idea: Last-click attribution over-credits channels that harvest existing demand and under-credits those that create it, so optimizing against it drains the demand it depends on.

Attribution models and their shared flaw

Attribution assigns credit for a conversion across the touchpoints that preceded it. Multi-touch models were developed to fix last-click by spreading the credit around.

Model	How credit is assigned	Systematic bias
Last click	All to the final click	Favors demand harvesting, especially branded search
First click	All to the first touch	Favors broad awareness channels regardless of effect
Linear	Equal across all touches	Treats a banner glimpse as equal to a store visit
Time decay	More to touches nearer the purchase	Assumes recency equals causation
Position based	Weighted to first and last	Arbitrary weights chosen by the analyst
Data driven	Modelled from observed paths	Still observational, and blind to unobserved paths

Now the point that matters more than the table. Every one of these models shares the same fundamental flaw, and rearranging the weights does not touch it. All of them assign credit among the touchpoints that were observed, and none of them asks the only question that matters: would this person have bought anyway? A model can distribute one hundred percent of the credit with great sophistication across five touchpoints and still be describing a purchase that would have happened with zero of them.

Three further problems compound this. Attribution sees only what it can track, which after the privacy changes in the last lesson is a shrinking and non-random subset of reality, and it is largely blind to offline media, word of mouth, and everything happening on other devices. View-through attribution, which credits an impression that was merely served and never clicked, is especially prone to abuse, because serving impressions is cheap and a long view-through window will eventually claim credit for almost everything. And the whole apparatus is a correlational description of paths, not a causal estimate, however precise the decimal places look.

Key idea: All attribution models distribute credit among observed touchpoints without ever asking whether the purchase would have happened anyway, which makes them descriptive rather than causal.

Marketing mix modelling

Marketing mix modelling takes the opposite approach. Instead of following individuals, it uses aggregate time-series data, typically weekly sales alongside spend by channel, price, distribution, promotions, seasonality, weather, and competitor activity, and fits a statistical model estimating each factor's contribution.

Its strengths are real and currently in fashion again. It needs no individual-level tracking, so privacy changes do not degrade it. It covers every channel including offline media, retail, and price. And it captures long-term and carryover effects through adstock terms rather than pretending advertising acts instantly.

Its weaknesses are equally real and are frequently understated by its vendors. It is correlational, and marketing variables move together: budgets rise before holidays, channels launch simultaneously, promotions coincide with campaigns, so collinearity makes separating their contributions genuinely difficult. It needs years of data with real variation in spend, and a brand that has spent the same amount in the same channels every quarter has given the model nothing to learn from. It is sensitive to specification, meaning two competent analysts can produce materially different answers from the same data. And it is slow, delivering conclusions quarterly at best, which is a poor fit for weekly decisions.

The professional way to use it is as a strategic allocation tool answering roughly how much should go where, not as a tactical instrument. And the most valuable practice, which good practitioners insist on, is to validate the model against experimental results wherever they exist, because a mix model calibrated by an incrementality test is far more trustworthy than one calibrated by the analyst's priors.

Key idea: Marketing mix modelling covers all channels and survives privacy changes but is correlational, collinear, specification-sensitive, and slow, so it belongs in strategic allocation and should be validated against experiments.

Incrementality

Incrementality is the question everything else is a proxy for: how many conversions happened because of the advertising that would not have happened without it? It is answered by creating a counterfactual, which in practice means withholding advertising from a randomly chosen group and comparing.

Four designs do most of the work. A randomized holdout withholds advertising from a random share of the addressable audience and compares outcomes; it is the cleanest available design where the platform supports it. Ghost or placebo advertising records which users would have been shown an advertisement and then serves something else, which gives an unusually clean control because the control group was selected by the same targeting machinery. Public service announcement controls serve an unrelated charity advertisement to the control group, achieving something similar. And geographic experiments split regions into matched treatment and control markets, which is the standard approach for offline and broad-reach media, and which requires careful matching and enough regions to be statistically meaningful.

Now the crucial reframing. In Lesson 7 you met eBay's finding of near-zero incremental return from branded search, and the Facebook comparison showing observational methods off by multiples. Both were incrementality tests contradicting attribution reports, and neither result would have been visible under any attribution model, because the attributed conversions were entirely real. They just were not caused. Attribution answers who was present at the sale; incrementality answers who caused it. Those are different questions, and only the second one carries a budget decision.

Incrementality has costs, and honesty requires naming them. A holdout is deliberately foregone revenue, which finance departments dislike. Tests take weeks and need scale to detect the small effects Lewis and Rao described. And they answer one question at one moment for one campaign rather than producing a standing dashboard. Those costs are exactly why the industry keeps preferring cheaper methods that produce more confident numbers.

Key idea: Attribution answers who was present at the sale and incrementality answers who caused it; only the second supports a budget decision, and it requires deliberately withholding advertising from a random control.

Triangulation and the metrics hierarchy

The mature answer is that no single method is sufficient and the three should be used together. Marketing mix modelling sets the strategic allocation across channels. Incrementality experiments calibrate and validate the model and settle specific disputes. Attribution is retained as a fast, directional, tactical signal, useful for spotting that something changed, and explicitly not treated as truth. That arrangement, sometimes called triangulation or unified measurement, is now the standard recommendation from serious practitioners, and its main obstacle is organizational rather than technical: it requires admitting that the precise number on last quarter's dashboard was not what it appeared to be.

Underneath method choice sits a simpler discipline that catches a great deal of nonsense. Sort every metric into three levels. Outputs are what you did: impressions served, posts published, press releases issued, reach delivered. Outcomes are what changed in people: awareness, recall, consideration, brand associations, intent. Impact is what changed in the business or the world: sales, market share, subscriptions, applications, behavior change. Most reporting in this industry consists of outputs presented in a tone appropriate to impact.

Public relations has fought this battle explicitly and largely won it in principle. Advertising value equivalency, the old practice of valuing coverage by what the equivalent advertising space would have cost, was a genuinely indefensible measure: it treats a hostile article as a positive because it was large, and it prices earned media using paid media rates for no reason other than convenience. The Barcelona Principles, developed by AMEC and revised repeatedly since 2010, reject it outright and insist on measuring outcomes and impact against stated objectives rather than counting clippings. Any practitioner still reporting advertising value equivalency in a formal setting is telling you something about their standards.

Keep one last principle in view, usually stated as Goodhart's law: when a measure becomes a target, it ceases to be a good measure. Optimize for clicks and you will get clicks, from people who click. Optimize for cheap impressions and you will get made-for-advertising sites. Optimize for attributed conversions and you will get advertising served to people who were already going to buy. Every measurement failure in this module has that shape, and the defense is to keep asking what the number was originally a proxy for.

Key idea: Triangulate mix modelling, experiments, and attribution rather than trusting one; sort metrics into outputs, outcomes, and impact; and remember that a measure optimized as a target stops measuring what it stood for.

Common misconceptions

• Multi-touch attribution fixes last-click's problem. It redistributes credit among observed touchpoints without ever asking whether the purchase would have happened anyway.
• More granular data produces more accurate measurement. Gordon and colleagues found rich individual-level data still failed to reproduce experimental benchmarks.
• Marketing mix modelling is objective because it is statistical. It is correlational and specification-sensitive, and two competent analysts can reach materially different conclusions.
• Advertising value equivalency measures public relations results. It prices earned coverage at paid rates and counts hostile articles as positives; the Barcelona Principles reject it.

Recap

• Last-click credits the final touch, systematically favoring demand harvesting over demand creation.
• All attribution models share one flaw: they never ask whether the purchase would have happened anyway.
• Marketing mix modelling covers everything and survives privacy changes, but is slow, collinear, and specification-sensitive.
• Incrementality requires a randomized control, delivered by holdouts, ghost advertising, PSA controls, or geographic splits.
• Triangulation uses mix modelling for allocation, experiments for validation, and attribution as a directional signal only.
• Sort metrics into outputs, outcomes, and impact, and reject advertising value equivalency.

Sources

1. Lewis, R. A., and Rao, J. M. (2015). The unfavorable economics of measuring the returns to advertising. Quarterly Journal of Economics, 130(4), 1941-1973. doi.org
2. Gordon, B. R., Zettelmeyer, F., Bhargava, N., and Chapsky, D. (2019). A comparison of approaches to advertising measurement: Evidence from big field experiments at Facebook. Marketing Science, 38(2), 193-225. doi.org
3. International Association for the Measurement and Evaluation of Communication. (n.d.). The Barcelona Principles. AMEC. amecorg.com
4. Institute for Public Relations. (n.d.). Measurement and evaluation research. instituteforpr.org
5. Wikipedia contributors. (n.d.). Attribution (marketing). Wikipedia. en.wikipedia.org

The big picture

A reporter covering technology for a national outlet receives somewhere between one and three hundred pitches on an ordinary day and writes perhaps one story. Do the arithmetic and the whole discipline of media relations comes into focus. Your pitch is not competing against silence. It is competing against two hundred other pitches, against the story the reporter is already writing, and against the reporter's entirely reasonable preference to be left alone.

That imbalance has become sharper over time, and the numbers are worth knowing. Pew Research Center analyses have documented a long decline in United States newsroom employment, with newspaper newsrooms losing a very large share of their staff since the mid 2000s, while the public relations workforce has grown. Fewer journalists, each covering more, receiving more pitches from more practitioners. If your mental model of media relations came from films, replace it with this one.

Media relations is the visible part of public relations, but it is only one of five practice areas, and this lesson covers the others too: internal communication, investor relations, public affairs, and community relations. They share a logic. Each identifies a public, learns what that public needs, and builds a relationship durable enough to survive the day you have bad news.

Key idea: Media relations is a competition for a shrinking amount of journalistic attention, so the discipline is about being genuinely useful to a specific reporter rather than about broadcasting announcements.

What journalists actually want

Start with news values, the criteria newsrooms use, consciously or not, to decide what is a story. Timeliness: is it happening now. Impact: how many people does it affect and how much. Prominence: does it involve someone or something already known. Proximity: is it near the audience, geographically or by interest. Conflict: is there a genuine disagreement. Novelty: is it unusual. Human interest: is there a person in it whose situation people will feel.

Read those again as a diagnostic. A product update with no new capability has none of them. The same update, framed around the people whose work it changes, with data showing scale, and released at a moment when the issue is live, may have four. The professional skill is not persuading a reporter that a non-story is a story. It is finding which real story you are actually sitting on, and quite often it is not the one the client wants to talk about.

The press release still exists and is far less central than students expect. Its structure remains useful discipline: a headline that states the news, a first paragraph carrying who, what, when, where, and why, then detail in descending order of importance in the inverted pyramid, then quotations that say something rather than praising the company, then boilerplate and a named contact. Writing one forces you to discover whether you have news. But most placements now come from a direct approach to a specific journalist, and a release is closer to a reference document than a pitch.

Key idea: News values are the newsroom's actual filter, and the skill is finding the real story you are sitting on rather than dressing up an announcement.

The pitch, and the rules of attribution

A good pitch is short, specific, and addressed to one person for a reason. Four things do most of the work. A subject line that states the story rather than the company name. A first sentence that says why this reporter, referencing something they have actually written, which takes two minutes and separates you from ninety percent of what lands in their inbox. The story in two or three sentences, including why now. And an offer of something concrete: an interview with a named person, data nobody else has, early access, a customer willing to talk.

Exclusivity is the main currency and it must be handled honestly. An exclusive offers a story to one outlet only, which raises the chance of substantial coverage and costs you everyone else. An embargo offers material to several outlets in advance on the agreed condition that nothing publishes before a stated time, which lets journalists prepare properly. Both depend entirely on your keeping the agreement, and a practitioner who breaks an embargo or quietly offers the same exclusive twice has ended relationships that took years to build.

The attribution conventions must be learned exactly, because misusing them causes real damage and reveals inexperience instantly. On the record means everything said may be quoted and attributed by name, and it is the default unless agreed otherwise in advance. On background usually means the information may be used and attributed to a described source, such as a company official, but not by name. Deep background typically means the information may be used with no source description. Off the record means the information may not be published at all, and it is properly used to help a reporter understand context, not to leak. Two rules follow. Terms must be agreed before the statement is made, never after. And they are not universal, so a careful practitioner states plainly what they mean rather than assuming.

Key idea: Pitch one reporter with a specific reason, treat exclusives and embargoes as binding, and agree attribution terms before speaking rather than afterward.

Internal communication

Here is a rule that sounds like a slogan and is a practical truth: your employees are your first public, and there is no such thing as an internal message. Anything sent to more than about fifty people is effectively public, because someone will screenshot it. Organizations that write internal communications they would be embarrassed to see published are creating a story, and the gap between the internal and external version is itself the story a reporter will write.

The functional case is stronger than the defensive one. Employees are the most credible source about an organization; the Edelman Trust Barometer has repeatedly found that people trust a company's employees more than its chief executive or its advertising. They are also the delivery mechanism for most of what an organization actually promises, since a service brand's advertising is a set of claims that frontline staff either honor or do not. And in a crisis, staff who learn about a layoff, a recall, or an investigation from the news rather than from their employer become a hostile public at exactly the wrong moment.

The practical rules are few and consistently violated. Tell employees first, or at worst simultaneously. Say what you know, what you do not know, and when you will next say something. Explain reasoning rather than issuing conclusions, because adults given only conclusions supply their own reasoning and it is usually worse than yours. And never use euphemism for serious things: describing layoffs as a realignment or a rightsizing does not soften the news, it teaches everyone that official language cannot be trusted, which costs far more later.

Key idea: Employees are the most credible and most consequential public, nothing internal stays internal, and euphemism for serious news destroys the credibility of every future message.

Investor relations and public affairs

Two specialisms operate under legal constraints tighter than the rest of the field, and both punish improvisation.

Investor relations communicates with shareholders, analysts, and the wider market, and in the United States it operates under securities law enforced by the Securities and Exchange Commission. The governing idea is materiality: information a reasonable investor would consider important in deciding whether to buy or sell. Regulation Fair Disclosure, adopted in 2000, prohibits selective disclosure of material non-public information to favored analysts or investors, requiring broad public dissemination instead, which is why earnings are announced by public release and open call rather than in private briefings. Add quiet periods before results, careful treatment of forward-looking statements, and the fact that a careless sentence can move a share price and attract an enforcement action, and you have a discipline where the communications instinct to be helpful must be subordinated to the legal requirement to be simultaneous and accurate.

Public affairs, or government relations, seeks to influence policy. Legitimate practice includes providing technical evidence to legislators and regulators, participating in formal consultations, filing comments on proposed rules, coalition building with genuine allies, and mobilizing real supporters. It is transparent and it is disclosed: federal lobbying activity in the United States must be registered and reported under the Lobbying Disclosure Act, and work on behalf of foreign governments and parties falls under the Foreign Agents Registration Act administered by the Department of Justice.

The bright line in public affairs is disclosure of who is speaking. Genuine grassroots advocacy, in which real people with a real stake contact their representatives, is a legitimate and valuable part of democratic practice. Astroturfing, in which an organization manufactures the appearance of that support through front groups, paid participants, or fabricated correspondence, is deception about the identity of the speaker. That is the same line Lesson 1 drew, it is prohibited by the PRSA code, and the next lesson takes it apart in detail.

Key idea: Investor relations is governed by materiality and the fair disclosure rule requiring simultaneous public dissemination, while public affairs is legitimate when registered and honest about who is speaking.

Community relations and the shape of a program

Community relations manages the relationship with the people who live near an operation, and it exists because organizations need what is sometimes called a social licence to operate: the informal consent of neighbors, without which planning permission, hiring, and daily operation become far harder. It is the least glamorous practice area and the one where the gap between real and performative work is widest, since a community can tell the difference between a company that consults before deciding and one that announces afterward with refreshments.

Whatever the practice area, a program has the same skeleton, usually taught as research, objectives, strategy, tactics, and evaluation. Research establishes the current position, and it is the step most often skipped. Objectives must be specific, measurable, and tied to a change in a public rather than to activity, which means raise favorability among state legislators from a measured baseline by a stated date, not build awareness. Strategy is the approach chosen; tactics are the things you make. Evaluation measures against the objectives you actually set, using the outcome and impact framework from the last lesson rather than counting clippings.

One honest caution about the whole discipline. Communication cannot fix a bad organization, and attempting it is where public relations earns its poor reputation. If a product is dangerous, a workplace is abusive, or a claim is false, the communications answer is to tell the organization to change the thing, and a practitioner whose counsel is only ever about presentation has stopped doing the job the professional codes describe.

Key idea: Every program runs research, objectives, strategy, tactics, evaluation, and no communication program can fix a problem that is actually about what the organization does.

Common misconceptions

• Public relations means sending press releases. Most coverage now comes from a specific pitch to a specific journalist; the release is closer to a reference document.
• Off the record means do not quote me by name. That is background; off the record means the information may not be published at all, and the terms must be agreed before speaking.
• Internal communications stay internal. Anything sent to more than a few dozen people is effectively public, and the internal version becoming public is itself the story.
• All grassroots campaigns are equivalent. Real supporters contacting representatives is legitimate advocacy; manufacturing that appearance through front groups is astroturfing and is prohibited.

Recap

• Newsroom employment has fallen sharply while the practitioner workforce has grown, intensifying competition for attention.
• News values, timeliness, impact, prominence, proximity, conflict, novelty, and human interest, are the newsroom's real filter.
• Pitch one reporter for a reason, and treat exclusives, embargoes, and attribution terms as binding.
• Employees are the most credible public, and euphemism for serious news destroys future credibility.
• Regulation Fair Disclosure requires material information to be released to everyone simultaneously.
• Lobbying is legitimate when registered and disclosed; manufacturing fake support is not.

Sources

1. Public Relations Society of America. (n.d.). About public relations. PRSA. prsa.org
2. Pew Research Center. (n.d.). Journalism and media research. pewresearch.org
3. U.S. Securities and Exchange Commission. (n.d.). Rules and regulations for public companies. sec.gov
4. U.S. Senate. (n.d.). Lobbying Disclosure Act reporting. lda.senate.gov
5. U.S. Department of Justice. (n.d.). Foreign Agents Registration Act. justice.gov

The big picture

A crisis is not simply bad news. It is an event that threatens an organization's ability to operate, that arrives faster than the organization's normal decision-making, and that is visible to people whose opinion matters. Those three features together explain why crises break organizations that handle ordinary bad news competently: the machinery that works at the speed of a weekly meeting fails at the speed of a video spreading on a Tuesday afternoon.

Here is the uncomfortable central fact of this lesson. Almost every serious failure of crisis communication is a failure that happened before the crisis. The plan that did not exist, the spokesperson who was never trained, the escalation path nobody had agreed, the culture in which raising a problem was career-limiting, the underlying operational fault that was known and tolerated. By the time the phone rings, most of the outcome has already been determined by decisions made calmly, months earlier, by people who were not thinking about a crisis.

We will work through the theory that tells you what response fits which situation, the preparation that makes a competent response possible, the specific demands of the first hours, and then three real cases studied carefully. I will treat all three factually, including the parts that complicate the tidy version usually taught.

Key idea: Most crisis outcomes are determined before the crisis, by preparation, escalation paths, and whether the organization had a culture in which problems could be raised.

Matching the response to the situation

W. Timothy Coombs's situational crisis communication theory gives the field its most useful framework. Its logic is that the right response depends on how much responsibility stakeholders attribute to the organization, because a response that does not match the attribution will fail regardless of how well it is written.

Coombs sorts crises into three clusters by attributed responsibility. The victim cluster covers situations where the organization is itself a victim, such as natural disasters, rumors, workplace violence, and product tampering; attribution is low. The accidental cluster covers technical failures and challenges where harm was unintended; attribution is minimal but real. The preventable cluster covers human error, knowing violations, and organizational misdeeds; attribution is strong, and this is where reputations are actually lost.

Cluster	Attributed responsibility	Example	Fitting response
Victim	Low	Product tampering, natural disaster, rumor	Inform, express concern, correct falsehoods
Accidental	Minimal	Technical-error accident, product recall	Explain, take action, avoid excuse-making
Preventable	Strong	Known defect ignored, misconduct, cover-up	Full apology, compensation, demonstrated change

The response strategies run from denial, through diminishment, which means excuse or justification, to rebuilding, which means apology and compensation, with bolstering as a supplement that reminds people of past good works. The theory's practical warning is about mismatch. Using denial or diminishment when responsibility is strongly attributed reads as evasion and makes things worse; that mismatch is the single most common failure pattern in real cases. The opposite error, over-apologizing for something you did not do, is less damaging reputationally but carries legal and precedent risks that counsel will raise.

One more finding worth carrying. Crisis history matters. An organization with prior similar incidents will have responsibility attributed more strongly for a new one, regardless of the facts of the present case, because audiences reason about patterns. A first incident is an accident; a third is a character.

Key idea: Match the response to attributed responsibility, because denial or excuse-making in the preventable cluster reads as evasion, and prior incidents raise the responsibility attributed to a new one.

Preparation

What preparation actually consists of is unglamorous and specific. A risk audit lists what can plausibly go wrong in this organization, ranked by likelihood and severity, and it should be written by people who know the operation rather than by the communications team alone. A crisis plan names who decides, who speaks, who is notified in what order, and how the team reaches each other outside working hours, because crises do not respect the calendar. Spokespeople are trained and rehearsed, since the ability to speak clearly under hostile questioning is a learned skill and not a personality trait.

Holding statements are drafted in advance for foreseeable scenarios, so the first public words are considered rather than improvised. A dark site, meaning a prepared web page that can be published immediately, gives the organization a place to put authoritative information that it controls. Monitoring must actually be watched, because organizations frequently learn about their own crises from a journalist. And stakeholders are mapped in advance: employees, customers, regulators, investors, partners, and communities each need something different, and working out that list during the crisis wastes the hours that matter most.

Underneath all of this sits the thing communications cannot supply. If the organization's culture punishes people who raise problems, the crisis team will learn about the defect at the same time as the public, and no plan survives that. The most valuable thing a senior communicator does is often not communication at all: it is insisting that a known problem be fixed while it is still a problem rather than a story.

Key idea: Preparation means a risk audit, a named decision chain, trained spokespeople, drafted holding statements, real monitoring, and a mapped stakeholder list, none of which can compensate for a culture that suppresses bad news.

The first hours

The United States Centers for Disease Control and Prevention's crisis and emergency risk communication guidance gives six principles that transfer well beyond public health: be first, be right, be credible, express empathy, promote action, and show respect. They are worth taking in order, because the sequence encodes real trade-offs.

Be first matters because the first credible account frames everything after it, and a vacuum will be filled by speculation, rumor, and whoever else is talking. Be right constrains be first: speed must not purchase inaccuracy, because a correction destroys more credibility than a delay. The resolution is the holding statement, which says what you know, what you do not know, what you are doing, and when you will speak again. That formula lets you be first without asserting anything you cannot support.

Express empathy comes before explanation, and getting this order wrong is the most common visible failure. When people have been hurt, an audience cannot hear your account of causes until you have acknowledged the harm, and an organization that leads with technical explanation or with its own difficulties sounds as though it has misunderstood what happened. Promote action tells people what to do, which reduces fear and restores agency, exactly as the efficacy research in Lesson 6 predicts. Show respect means treating the affected as people rather than as a reputational problem, and audiences detect the difference reliably.

Two further rules. Never speculate about cause while an investigation is running, because being wrong publicly is worse than saying you do not yet know. And accept that legal counsel and communications will disagree: counsel minimizes admissions to limit liability, communications urges acknowledgment to preserve trust. That tension is legitimate and it must be resolved by a decision maker senior to both, in advance, not argued out live while the story runs.

Key idea: Be first, be right, be credible, express empathy, promote action, show respect, and use a holding statement to be fast without asserting what you cannot yet support.

Tylenol, 1982

In the autumn of 1982 seven people in the Chicago area died after taking Extra-Strength Tylenol capsules that had been laced with cyanide. The tampering occurred after the product left Johnson and Johnson's control, placing the case squarely in Coombs's victim cluster. What made the response famous was that the company acted as though its responsibility were far greater than the attribution warranted.

Johnson and Johnson, under chief executive James Burke, warned the public and the medical community, halted production and advertising, and recalled roughly thirty one million bottles nationally at a cost widely estimated near one hundred million dollars in 1982 terms. It cooperated openly with police and the Food and Drug Administration, communicated continuously through the press rather than through lawyers, and then rebuilt the product with tamper-evident packaging, an innovation that became an industry standard and later a regulatory requirement. Tylenol returned to the market and recovered its share within roughly a year. The murders were never solved.

The case is taught as the model response, and mostly it deserves that. But you should know how the tidy version simplifies. The recall escalated over days rather than arriving fully formed in the first hour, and it was shaped by unfolding events, regulatory pressure, and the discovery of further contaminated bottles. The company had a strong pre-existing reputation and a widely cited corporate credo, which gave its actions credibility that a less trusted firm would not have enjoyed. And Johnson and Johnson has had subsequent episodes handled far less well, which is a reminder that a good response is a decision made under particular conditions rather than a permanent corporate character trait.

The transferable lessons are three. Put public safety ahead of short-term financial cost, and do it visibly and early. Communicate directly and continuously rather than through minimal legal statements. And fix the underlying vulnerability in a way people can see, because tamper-evident packaging was a communication as much as an engineering change.

Key idea: Tylenol's response worked because the company acted with more responsibility than the situation attributed to it, communicated continuously, and made the fix visible, though the tidy textbook version understates how much it escalated over days.

Two modern failures

The Deepwater Horizon disaster began on 20 April 2010 when the drilling rig exploded, killing eleven workers and beginning the largest marine oil spill in history. BP's communication compounded an operational catastrophe in three specific ways. It repeatedly issued flow-rate estimates far below what independent scientists calculated, so each correction made the company look either incompetent or dishonest. Its chief executive publicly minimized the environmental effect early on. And in a televised apology he added that he would like his life back, a sentence that placed his own inconvenience beside eleven deaths, followed weeks later by his attendance at a yacht race while the well was still flowing.

Read those failures against the framework. The crisis sat in the preventable cluster, where full acknowledgment is the fitting response, and the company reached for diminishment. It broke be right by asserting numbers it could not support. And it inverted the empathy sequence completely, foregrounding its own position when the audience needed acknowledgment of harm. No communications strategy could have made the spill acceptable, but the response measurably worsened an already severe position.

The second case is faster and cleaner as a teaching example. On 9 April 2017 a passenger was forcibly dragged from a United Express flight at Chicago O'Hare after being selected for removal from a full aircraft, and passenger video spread worldwide within hours. The airline's first statement apologized for having to re-accommodate customers, a euphemism so far from the visible footage that it became the story, and an internal message from the chief executive supporting the crew and describing the passenger in unflattering terms leaked immediately, precisely as the previous lesson warned. A fuller and more direct apology followed days later, along with policy changes and a settlement, but by then the first response had defined the event.

The lesson here is about the gap between language and evidence. In an environment where everyone has seen the video, a euphemism does not soften the event, it discredits the speaker, and it converts a bad incident into a bad incident plus proof that the organization will not describe reality accurately. Note also that this crisis moved from event to global story in hours, which is the operating speed you must now plan for.

Key idea: BP diminished a preventable crisis, asserted numbers it could not support, and centered itself over victims, while United's euphemism collided with video everyone had seen, discrediting the speaker and becoming the story.

Common misconceptions

• Crisis communication is about handling the press. Most of the outcome is decided by preparation and by whether the underlying problem gets fixed.
• Saying nothing until the facts are in is safest. A vacuum is filled by speculation; the holding statement lets you be first without asserting what you cannot support.
• An apology is always the right response. Situational crisis communication theory matches response to attributed responsibility, and over-apologizing carries legal and precedent risk.
• The Tylenol response was instantaneous and complete. The recall escalated over days under unfolding events and regulatory pressure, and the company had a strong prior reputation to draw on.

Recap

• A crisis threatens operations, moves faster than normal decision-making, and is visible to publics that matter.
• Coombs sorts crises into victim, accidental, and preventable clusters, and mismatching the response is the commonest failure.
• Preparation means a risk audit, decision chain, trained spokespeople, holding statements, monitoring, and stakeholder mapping.
• Be first, be right, be credible, express empathy, promote action, show respect, and never speculate on cause.
• Tylenol worked by exceeding its attributed responsibility, communicating continuously, and making the fix visible.
• BP diminished a preventable crisis and centered itself; United's euphemism collided with video everyone had seen.

Sources

1. Centers for Disease Control and Prevention. (n.d.). Crisis and emergency risk communication. CDC. emergency.cdc.gov
2. Institute for Public Relations. (n.d.). Crisis management and communications research. instituteforpr.org
3. Wikipedia contributors. (n.d.). Chicago Tylenol murders. Wikipedia. en.wikipedia.org
4. Encyclopaedia Britannica. (n.d.). Deepwater Horizon oil spill of 2010. britannica.com
5. Wikipedia contributors. (n.d.). United Express Flight 3411 incident. Wikipedia. en.wikipedia.org

The big picture

Everything in this course has been building toward one line, and you already have it from Lesson 1. Persuasion is legitimate. Deception about who is speaking, about whether they were paid, or about facts that can be checked is not. Nearly all of American advertising law is that sentence worked out in detail, and nearly all of the professional ethics is that sentence applied where the law does not reach.

The legal architecture is narrower than students expect. There is no licence to practice advertising or public relations, no board that can strike you off, and no pre-approval of most advertising. What exists instead is enforcement after the fact by the Federal Trade Commission, an industry self-regulatory layer, sector-specific regulators for food, drugs, financial products, and alcohol, and professional codes that bind only those who choose to join. That structure places an unusual amount of weight on individual judgment, which is why the ethics half of this lesson is not decoration.

We cover the FTC's core standards, the disclosure rules, the special protections around children and privacy, the two named abuses of greenwashing and astroturfing, the PRSA code, and finally how people actually enter these industries, because the ethics of this field are mostly decided by ordinary people early in their careers deciding what they will and will not do.

Key idea: There is no licence and almost no pre-approval in these industries, so after-the-fact enforcement, self-regulation, and individual judgment carry the weight.

The FTC's core standards

Section 5 of the Federal Trade Commission Act prohibits unfair or deceptive acts or practices, and the Wheeler-Lea Act of 1938 from Lesson 2 is what extended that to protect consumers directly. Three standards do the work.

Deception has three elements. There must be a representation, omission, or practice likely to mislead; the audience must be judged as a consumer acting reasonably in the circumstances; and the claim must be material, meaning likely to affect a purchase decision. Note what is absent: intent to deceive is not required. An advertiser who genuinely believed a false claim has still violated the standard, which is why the substantiation habit matters so much.

Substantiation requires that advertisers possess adequate support for objective claims before disseminating them. For health, safety, and efficacy claims the expected support is competent and reliable scientific evidence, which in practice usually means well-designed human studies rather than testimonials or a supplier's brochure. Unfairness, the third standard, reaches practices causing substantial consumer injury that consumers cannot reasonably avoid and that is not outweighed by benefits, and it covers conduct that is not strictly a false statement.

Two distinctions repay attention. Express claims are stated outright; implied claims are what a reasonable consumer takes away, and advertisers are responsible for both, so the defense that you never actually said it usually fails. And puffery, meaning obvious subjective exaggeration such as the best coffee in the world, is permitted because nobody treats it as a factual claim, while a specific measurable statement is not puffery however enthusiastically it is delivered. The line runs between opinion nobody could verify and assertion somebody could check.

Key idea: Deception requires only that a material claim is likely to mislead a reasonable consumer, with no intent needed, and advertisers are responsible for implied claims as well as express ones.

Disclosure: native advertising, endorsements, and reviews

Disclosure is where the modern enforcement action lives, because the modern problem is paid material dressed as unpaid material. The governing requirement is that disclosures be clear and conspicuous, which the Commission has consistently interpreted by effect rather than by form: the question is whether ordinary consumers actually notice and understand it, not whether the words exist somewhere on the page.

Native advertising is paid content designed to match the form and function of the surrounding editorial material. The FTC's guidance is that consumers should be able to tell it is advertising before they engage with it, which means the label sits before or at the beginning of the content, uses plain terms such as advertisement or sponsored rather than vague ones such as promoted partner or brought to you by, and does not rely on the audience recognizing a small logo. The theory is simple: the harm is not the advertising, it is the borrowed credibility of the publication.

The endorsement guides cover any material connection between an endorser and a brand, and material connection is defined broadly enough to include payment, free products, discounts, employment, and family relationships. Testimonials must reflect honest opinions and typical results, and the old device of showing an exceptional outcome with a small results not typical caption has been rejected: if the depicted result is not what consumers generally achieve, the advertiser must disclose what they generally do achieve. In 2024 the Commission finalized a rule addressing fake and manipulated consumer reviews and testimonials, which reaches purchased reviews, undisclosed insider reviews, and suppression of negative ones, with civil penalties available. Read that alongside Lesson 6: social proof is the most powerful and most easily faked of the influence principles, and this is the law catching up with it.

Key idea: Disclosure is judged by whether consumers actually notice and understand it, native advertising must be identifiable before engagement, and manufactured reviews are now separately actionable.

Children, privacy, and environmental claims

Three areas carry heightened rules, and each rests on a specific rationale worth knowing.

Children are protected because the developmental evidence is unambiguous. Young children do not reliably distinguish advertising from programming, and the understanding that advertising has persuasive intent and a commercial source develops later still, typically emerging around the ages of seven and eight and consolidating after that. A message aimed at someone who cannot recognize it as a message is not persuasion in any meaningful sense. The practical framework includes the Children's Online Privacy Protection Act, which requires verifiable parental consent before collecting personal information from children under thirteen; broadcast rules limiting commercial time in children's programming and prohibiting host-selling, where a program's own characters sell products within it; and the Children's Advertising Review Unit within BBB National Programs, which reviews advertising directed to children against its own guidelines.

Privacy has become advertising law because, as Lesson 11 showed, the bid request is a description of a person. The United States has no single federal privacy statute for advertising, so the field is governed by the FTC's unfairness and deception authority, sectoral laws covering health and financial data, and state statutes led by California's consumer privacy law, while the European Union's General Data Protection Regulation reaches any organization handling data about people in the EU. The recurring themes are notice, choice that is genuine rather than engineered, heightened treatment of sensitive categories such as health, precise location, and information about children, and enforcement against dark patterns, meaning interface designs that steer people into consent they did not intend to give.

Environmental claims are governed by the FTC's Green Guides, whose central principle is that broad unqualified claims such as eco-friendly, green, or sustainable are nearly impossible to substantiate and should be replaced by specific qualified claims about a named attribute. Greenwashing, meaning marketing that presents an organization as more environmentally responsible than its conduct supports, ranges from vague labelling to outright fraud; the Volkswagen emissions case, in which vehicles marketed on clean diesel performance contained software designed to detect and pass emissions tests, is the extreme end and produced criminal and civil consequences rather than merely reputational ones.

Key idea: Children's rules rest on the developmental fact that persuasive intent is not understood until around seven or eight, privacy is now advertising law because the bid request describes a person, and the Green Guides require specific qualified claims rather than broad environmental adjectives.

Astroturfing, named and rejected

Astroturfing is the manufacture of apparent grassroots support: front groups presented as independent citizen organizations, paid or fabricated commenters, sockpuppet accounts, template letters submitted as though individually written, and undisclosed funding of apparently independent experts or advocacy. The name is a joke about artificial grass, and the joke is the definition, because what is faked is authenticity itself.

It should be rejected without qualification, and it is worth being precise about why, because the reasoning generalizes. Astroturfing is deception about the identity of the speaker, which is the first of the three lines this course has used throughout. It is not made acceptable by the underlying position being defensible, since a good argument does not need a fake constituency and using one concedes that the real constituency is insufficient. It corrupts the mechanisms democracies use to gauge public opinion, which means the harm falls on people who never encountered the campaign at all. And it is specifically prohibited by professional codes: the PRSA Code of Ethics requires disclosure of the sponsors of causes and interests represented, and it is difficult to imagine a clearer breach.

The legitimate version of the same activity is straightforward and effective. Fund advocacy openly under your own name. Help real supporters contact their representatives in their own words. Disclose who paid for research you cite. If your position cannot attract genuine support once people know who is asking, that is information about your position rather than a communications problem to be engineered around.

Key idea: Astroturfing fakes authenticity itself, which is deception about the speaker's identity, and a defensible argument that needs a fabricated constituency is telling you something about its real support.

The professional codes

Because there is no licence, professional codes are voluntary and are enforced mainly through membership and reputation. The PRSA Code of Ethics is the field's most cited, and it is short enough to read in full, which you should do.

It sets out six professional values: advocacy, honesty, expertise, independence, loyalty, and fairness. Notice that advocacy and honesty sit together at the top, which is the code refusing the false choice students often assume; a practitioner is a responsible advocate, not a neutral reporter, and advocacy is legitimate precisely because it is honest and disclosed. The provisions of conduct then cover the free flow of accurate information, fair competition, disclosure of information including the sponsors of causes represented, safeguarding confidences, avoiding conflicts of interest, and enhancing the profession.

Two limitations should be stated plainly rather than glossed. Enforcement reaches only members and the maximum sanction is expulsion, so the code binds through professional identity rather than through law. And the values genuinely conflict in practice, most obviously loyalty to a client against honesty to the public, and the code cannot resolve those conflicts for you. What it does supply is a vocabulary for naming the conflict out loud, which is frequently the whole of what is needed in the meeting where the decision gets made.

Key idea: The PRSA code names advocacy and honesty as compatible values, requires disclosing the sponsors of causes represented, and binds through professional identity rather than law.

Careers

Entry routes differ by function and it helps to be concrete. Creative roles, copywriter and art director, are entered on the strength of a portfolio, still called a book, usually built in pairs and containing perhaps eight to twelve campaigns that demonstrate thinking rather than finished production values. Strategy and account planning reward evidence that you can turn research into an insight, which a written strategy document or a well-argued analysis of an existing campaign will show better than a degree. Account management is entered through internships and rewards organization, judgment, and the ability to be trusted with a difficult conversation. Media and analytics increasingly reward quantitative skill and comfort with data tools. Public relations entry is typically through internships and agency assistant roles, with writing quality the single most reliable differentiator.

The United States Bureau of Labor Statistics publishes occupational profiles for public relations specialists and for advertising, promotions, and marketing managers, covering typical duties, entry-level education, median pay, and projected growth, and it is the sensible place to check current figures rather than trusting an industry brochure. Student competitions such as the American Advertising Federation's National Student Advertising Competition produce genuine portfolio work, and the PRSA offers accreditation, the APR credential, for practitioners some years into a career.

Two closing pieces of advice that this course has earned the right to give. First, learn to measure honestly, because the field is short of people who can tell a client that the number on the dashboard does not mean what it appears to mean, and that skill compounds. Second, decide your lines before you are asked to cross them. Nobody is ever asked to do something obviously wrong on their first day; they are asked to do something slightly ambiguous, and then something slightly further, by people they like, under deadline. Having already decided what you will not do, in advance and in your own words, is the only defense that reliably works.

Key idea: Entry runs through portfolios, internships, and demonstrated writing or analytical skill, and the two durable advantages are measuring honestly and deciding your ethical lines before anyone asks you to cross them.

Common misconceptions

• An advertiser must intend to deceive to violate the law. Intent is not an element; a material claim likely to mislead a reasonable consumer is enough.
• Adding results not typical fixes an exceptional testimonial. The advertiser must disclose the results consumers generally achieve, not merely caption the exception.
• Astroturfing is acceptable if the underlying cause is good. It is deception about who is speaking, and a good argument that needs a fake constituency lacks a real one.
• Professional codes have legal force. They bind members through professional identity, with expulsion as the maximum sanction; there is no licence to lose.

Recap

• Deception requires a material claim likely to mislead a reasonable consumer, with no intent needed and implied claims included.
• Substantiation must exist before the claim runs, and health and efficacy claims need competent and reliable scientific evidence.
• Disclosure is judged by whether consumers actually notice; native advertising must be identifiable before engagement.
• Children's protections rest on persuasive intent developing around ages seven and eight; COPPA governs data from under-thirteens.
• The Green Guides require specific qualified environmental claims, and astroturfing is prohibited as deception about the speaker.
• The PRSA code pairs advocacy with honesty and requires disclosing the sponsors of causes represented.

Sources

1. Federal Trade Commission. (n.d.). Advertising and marketing guidance. ftc.gov
2. Federal Trade Commission. (n.d.). The FTC's endorsement guides: What people are asking. ftc.gov
3. Federal Trade Commission. (n.d.). Children's privacy and COPPA guidance. ftc.gov
4. Public Relations Society of America. (n.d.). PRSA Code of Ethics. prsa.org
5. U.S. Bureau of Labor Statistics. (n.d.). Public relations specialists. Occupational Outlook Handbook. bls.gov
6. Wikipedia contributors. (n.d.). Astroturfing. Wikipedia. en.wikipedia.org