You see Tom Cruise in an ad. You don't know why he's there. Maybe he genuinely loves the product — a real actor, choosing to put his name on something he actually uses. Or maybe the company paid him an enormous amount of money to be there, and he doesn't care about the product at all. Here's the thing Rory Sutherland points out: it doesn't matter which one it is. Both branches of that fork lead to the same place — the product is worth looking at.1
Most persuasion techniques work by making one specific claim credible. Celebrity endorsement, on this reading, works differently — it doesn't need you to believe any single claim. It offers you two mutually exclusive explanations for why a famous person is attached to a product, and both of the available explanations point toward the same conclusion. If the celebrity's involvement is sincere, that's a real, personal signal of quality — someone with more discernment and more alternatives than you have chosen this thing. If the celebrity's involvement is purely commercial, that still tells you something real: the company was willing to spend an enormous, verifiable sum of money on this specific message, and companies don't do that carelessly.2 The second branch is really a variant of costly signaling — see Costly Signaling — routed through a famous face instead of a full-page ad or a hand-delivered envelope.
A skeptical viewer who has learned to distrust celebrity endorsements generally — "of course they're paid, they don't actually care" — usually thinks this insight defeats the persuasion technique. It doesn't, on Sutherland's logic, because the technique never actually required the viewer to believe the endorsement was sincere. Realizing "they were probably just paid" doesn't dismantle the signal. It just moves you from branch one to branch two, and branch two is still evidence worth weighing: this company thought this specific message, delivered by this specific expensive person, was worth a serious sum of money. That's a real data point about the company's confidence in the product, independent of whether the celebrity personally believes a word of it.
The logic breaks down only when the amount spent no longer reads as meaningfully costly relative to the company's scale — a trivial expense for a company that size proves nothing, the same way a bank robber tipping a dollar proves nothing about generosity. It also breaks down when audiences develop specific knowledge that undercuts the "the company wouldn't do this carelessly" assumption — the moment a particular celebrity becomes known for endorsing literally anything for a fee, without any evident selectivity, the endorsement stops reading as evidence of anything except that the celebrity's agent answered the phone.
You're advising a mid-size brand considering a celebrity partnership, and the budget under discussion is modest relative to the celebrity's usual fee — enough to get their name attached, not enough to be a serious financial commitment for either party. You flag the risk directly: a partnership too cheap to register as a real expenditure doesn't trigger the second-order logic at all. It just looks like what it is, a modest transactional deal, and buyers who've seen a hundred of these won't run either branch of the inference — they'll assume, correctly, that nobody involved is taking much of a risk.
Later, the same brand lands a genuinely expensive, high-profile partnership, and the marketing team wants to keep the commercial arrangement quiet — worried that revealing how much was paid will make the endorsement look cynical. You argue the opposite: for this mechanism to work, the cost has to be legible, even if the sincerity isn't. A viewer who has no idea the arrangement was expensive can't run the "they wouldn't spend that carelessly" branch of the logic. Some visibility into the scale of the investment — not necessarily the exact number, but a sense that this was a serious commitment — is doing real persuasive work, and hiding it entirely may be giving away half the mechanism's power.
The clearest evidence for this dual-branch structure is that celebrity endorsement continues to work, reliably, even in a media culture that is thoroughly cynical about the sincerity of paid endorsements — audiences broadly assume most endorsements are paid, and the technique still moves product, which is hard to explain if the mechanism depended entirely on believed sincerity.
The open tension: the source presents this as a clean two-branch logic without specifying how a viewer actually weighs the two branches against each other, or what determines which branch a given viewer defaults to. Some viewers may be highly cynical (default to branch two, "they were paid," and weight it less than a viewer who defaults to branch one). The claim that "either way you should investigate" treats both branches as roughly equally persuasive, which may not hold for every audience or every category of product.
This page's account sits comfortably alongside Social Proof Fabrication built earlier in this same batch, but the two describe importantly different failure modes for the same broad category of technique. Fabricated social proof fails when the audience discovers the proof was never real at all — fake reviews, staged testimonials — which destroys the signal entirely once exposed. Celebrity endorsement's second-order logic is more robust precisely because it doesn't require the "proof" (the celebrity's apparent enthusiasm) to be real; it has a fallback interpretation built in. The convergence: both techniques exploit an audience's difficulty distinguishing genuine signals from manufactured ones. The tension: only one of them has a legitimate fallback that survives the audience figuring out the manufacture.
Behavioral-Mechanics — Costly Signaling. Costly signaling explains why spending money credibly proves conviction; this page shows a specific delivery mechanism for that general principle — routing the cost through a recognizable human face rather than through raw ad spend, packaging, or waste. The insight the pairing produces: celebrity endorsement is not really a distinct persuasion category from costly signaling in general, it's costly signaling with a face attached, and the face's main function is making an otherwise-invisible expenditure (how much did this campaign actually cost?) instantly visible and memorable to an audience that would never see a media-buy invoice.
Behavioral-Mechanics — Social Proof Fabrication. Both techniques exploit the same underlying audience vulnerability — difficulty distinguishing a genuine signal of quality from a manufactured one — but this page's mechanism is structurally more resilient because it doesn't require the audience to be fooled about sincerity to still work. The insight the pairing produces: persuasion techniques that build in a legitimate fallback interpretation (celebrity endorsement's "either way, worth investigating") survive audience skepticism in a way that techniques dependent on a single, falsifiable claim (fabricated social proof's "everyone loves this") do not — a general principle about which manipulation techniques age well and which get defeated the first time an audience catches on.
Sharpest implication: the most durable persuasion techniques are not the ones that make the most convincing single claim — they're the ones built so that even a skeptical, correct rejection of the surface claim still lands the audience on a conclusion the advertiser wanted.
Generative questions: