The Cost of Being Believed: Why Effortless Marketing Stopped Persuading
The most expensive thing a brand can do right now is look effortless
In November 2025, Coca-Cola released its second AI-generated holiday campaign in as many years. The animation was smoother than the 2024 version. The lighting was cleaner. The red trucks gleamed. By every technical measure, the work had improved. And audiences hated it anyway — calling it “soulless” and “devoid of any actual creativity,” with some threatening to switch to Pepsi over a Christmas commercial.
The instinct is to read this as a story about taste, or nostalgia, or a brand mishandling a beloved franchise. It is none of those things. It is a story about economics. Coca-Cola broke the oldest rule in persuasion: it stopped paying for the thing it was asking people to believe.
Polish used to be proof. That link is now severed.
For a century, a beautifully produced advertisement carried information that had nothing to do with its content. A thirty-second film shot on location, scored, and cut to broadcast standard told the viewer something the script never said out loud: this company spent a fortune to reach you. And a company that spends a fortune is a company that expects to be around long enough to earn it back. The gloss was never just decoration. It was collateral.
This is why television advertising still outperforms cheaper channels on nearly every measure of brand trust — financial strength, quality, popularity — even in a landscape where attention has fragmented across a hundred smaller screens. Thinkbox’s research isn’t measuring the persuasiveness of the message. It’s measuring the persuasiveness of the expense.
Generative AI severed that link. When a finished, polished, broadcast-grade asset can be produced in an afternoon for the cost of a subscription, polish stops carrying any information at all. The gloss is still there. The collateral behind it is gone. And audiences — who read these signals emotionally, long before they reason about them — feel the absence immediately, even when they can’t name it.
The gazelle that jumps in front of the lion
The framework that explains all of this predates marketing by millions of years. In 1975, the evolutionary biologist Amotz Zahavi proposed what he called the handicap principle: the most trustworthy signals in nature are the ones that are expensive to produce and impossible to fake.
His favorite example was stotting — the behavior where a gazelle, on spotting a predator, springs straight up into the air instead of fleeing. It looks suicidal. It wastes precious time and energy. That is exactly the point. Only a genuinely fast, healthy gazelle can afford to burn resources showing off, so the display is an honest advertisement of fitness. A weak animal cannot fake it. The cost is the message.
Costly signaling theory has since become one of the most durable ideas in behavioral science, and it maps onto brand persuasion with uncomfortable precision. Robert Cialdini’s principle of Authority describes the same mechanism from the receiving end: we defer to signals of competence and standing. Zahavi explains why those signals work only when they cost something to send. A luxury watchmaker’s decade-long waitlist, a law firm’s marble lobby, a brand’s willingness to run a single ad during one football game for the price of a house — these are all forms of stotting. The waste is not a bug. It is the entire argument.
Which means the question every marketer now faces is not “how do we make our content better.” It is “what are we still willing to spend that our competitors cannot cheaply imitate.”
What actually collapsed
The data on AI content is not subtle. Consumer preference for AI-generated creator content fell from roughly 60% in 2023 to 26% by 2025 — a collapse in trust that tracked almost exactly with the technology getting good enough to flood every feed. Abundance did not build familiarity. It bred suspicion.
A 2025 study from the Nuremberg Institute for Market Decisions found that merely labeling an ad as AI-generated made people rate it as less natural and less useful, which in turn lowered their attitude toward the ad and their willingness to research or buy. Note what that finding isolates: nothing about the ad changed except the disclosure. The pixels were identical. What shifted was the audience’s read on what the ad cost to make — and with it, their willingness to be moved by it.
This is persuasion knowledge activating in real time. The moment a viewer suspects a message was cheap to produce, they stop treating it as evidence and start treating it as noise. The content can be flawless and still fail, because the failure is not aesthetic. It is structural.
The “made by humans” badge is a signal — but a fragile one
Brands have noticed, and a countermovement is already underway. When Aerie pledged not to use AI in its advertising, the post became its most-liked in a year — over 40,000 likes and hundreds of comments. Dove committed to never using AI to represent real women. Le Creuset publicly annotated a piece of content to confirm it was made by hand. iHeartMedia rolled out a “guaranteed human” tagline after finding that 90% of its listeners wanted media made by people. One trade analyst has started calling the “no AI” badge the new organic label.
The instinct is right. The execution is where most brands will get it wrong. Because a claim of humanity is not, by itself, a costly signal. Saying “this was made by a person” is free. Anyone can type it, including the brands generating everything else with AI behind the curtain. The badge only works if it is backed by something a pretender cannot afford to fake.
Dove’s pledge carries weight because Dove has spent twenty years and enormous sums building the Real Beauty platform — the promise is expensive precisely because breaking it would torch two decades of investment. Aerie’s lands because the brand’s entire positioning has been unretouched bodies since 2014. The signal is credible in proportion to what the brand would lose by being caught lying. Slap “human-made” on a product with no such history and you have not sent a costly signal. You have sent a coupon.
The same logic governs earned media, which is why public relations still outperforms paid content on trust. A journalist’s independent coverage persuades precisely because you cannot simply buy it — it carries the cost of someone else’s reputation vouching for you. The moment that coverage looks purchased, planted, or machine-written, it collapses back into an advertisement and loses the premium. Third-party credibility is costly signaling outsourced: the expense is borne by the source, and the audience prices it accordingly. As generative tools make it trivial to manufacture the appearance of consensus — fake reviews, synthetic testimonials, AI-spun “coverage” — the signals that survive are the ones with a real name and a real reputation staked behind them.
Engineering signals that cost something
The strategic move is not to chase authenticity as a mood. Authenticity is a feeling audiences have; it is not a lever you can pull. The lever is cost — deliberately building signals into your marketing that are expensive, verifiable, and hard to counterfeit. A few forms this takes:
Specificity over polish. A vague, glossy claim is cheap. A precise one — a named customer, an exact figure, a dated result, a flaw admitted openly — is expensive, because it can be checked and therefore can be wrong. Specificity is costly signaling in prose. It exposes you, which is exactly why it persuades.
Provenance you can prove. Show the work. The founder on camera without a script. The factory floor. The research methodology. The raw footage alongside the finished cut. Provenance is stotting: it burns the resource of exposure to demonstrate you have nothing to hide.
Commitments with teeth. A guarantee that would genuinely hurt to honor signals confidence that a soft promise cannot. Cialdini’s principle of Commitment works on your audience; a costly commitment works because it works on you first.
Scarcity that reflects real constraint. Manufactured urgency is now trivially cheap and universally distrusted. Scarcity earns belief only when it maps to an actual limit — a real waitlist, a genuinely capped cohort, a thing that is hard to get because it is hard to make.
Each of these has the same structure as the gazelle’s jump. It costs the sender something, and that cost is the reason the receiver believes it.
The quiet advantage
There is an irony worth sitting with. The flood of frictionless content was supposed to level the field — to let anyone produce agency-grade work for nothing. Instead it did the opposite. When everyone can produce the polished thing for free, the polished thing signals nothing, and the advantage returns to whoever is still willing to pay a cost their competitors will not.
That cost does not have to be money. It can be exposure, precision, restraint, the discipline to say less and mean it, the patience to build something slowly enough that it cannot be faked overnight. These have always been the expensive materials. The AI era has simply made them scarce again, and scarcity is where persuasion lives.
Coca-Cola’s mistake was not that it used a new tool. It was that it tried to buy belief without paying for it — and audiences, who have been reading the honesty of signals since before we had language, priced it instantly. Influence has never been about looking effortless. It is about being willing to spend what a pretender won’t.
