Proof Is Now the Scarcest Signal in Marketing

A signal is only informative if it is costly to fake. This is not a marketing idea; it is a result from evolutionary biology and it holds everywhere. A peacock’s tail conveys information precisely because a weak peacock could not carry one. The moment tails become free, tails stop meaning anything, and the peahens have to find something else to look at.

Marketing has just had its tails made free.

The signals that stopped working

Consider what a buyer has traditionally used to assess a marketing partner. A well-argued strategy document. A confident, articulate proposal. A beautifully designed case study. A website that reads as though written by people who know what they are talking about. A thought-leadership essay, in fact, rather like this one.

Every single item on that list can now be produced, to a genuinely high standard, by an organisation with no clients, no track record, and no competence whatsoever, in less time than it takes to read it. The cost of producing the signal has collapsed to nearly zero, and so — necessarily, immediately, and permanently — the signal has stopped carrying information.

This is worth sitting with, because it invalidates most of how the industry sells itself. The pitch document is dead as evidence. It may still be required as ritual, but nobody should mistake it for information any longer.

A claim that anyone can generate is not evidence. It is decoration that happens to be made of words.

The temptation, and why it is fatal

There is an obvious and disastrous response to this, and a great deal of the market is currently taking it: if the signal is cheap, produce more of it. Flood the zone. More content, more case studies, more claims, more numbers.

The numbers are where this becomes career-ending. It is trivially easy to write “we drove a 340% increase in engagement for a global beverage brand”, and such sentences are now everywhere. They are almost always unverifiable, frequently unattributable, and sometimes simply invented — a percentage chosen because it sounds impressive but not so impressive as to be unbelievable. We have seen this done. We regard it as the single fastest way to destroy a firm’s credibility, and it is destroyed permanently, because the buyer who catches one fabricated number correctly assumes there are others.

The asymmetry is brutal. A hundred true claims buy you a little credibility. One false claim removes all of it.

What proof actually looks like

Proof is not a number. Proof is a number with a method attached that a hostile reader could interrogate. The distinction is the whole thing.

• A stated baseline. Compared to what? Over what period? A result without a baseline is a sentence, not a finding.
• A control. What did the untreated group do? Without a holdout, you have not measured your impact; you have measured the weather and taken credit for the sunshine.
• A named mechanism. Why would this have worked? A result with no plausible mechanism is usually a measurement artefact wearing a costume.
• The failures, included. A partner who reports only wins is not reporting. They are selling, and you are reading a brochure.

Almost no marketing case study in circulation meets this standard. That is not an accident and it is not merely laziness — meeting it is expensive, it is slow, and it produces less flattering results than not meeting it. Which is precisely why it is now the only credible thing left. Proof is valuable because it is costly. That is the entire mechanism.

The corollary for AI search

There is a second reason this matters now, and it is mechanical rather than philosophical.

Buyers increasingly form their first impression of a firm through an AI system that has summarised it. Those systems are trained on the whole web, and they weight corroboration: a claim a company makes about itself carries far less weight than the same claim made about it somewhere else. A firm whose evidence exists only as adjectives on its own homepage is, from the model’s perspective, a firm about which nothing is known.

So the incentive now runs in the same direction as the ethics, which is a rare and pleasant alignment. Documented, corroborated, externally verifiable outcomes are simultaneously the honest thing to publish and the thing most likely to be surfaced by the systems that increasingly mediate discovery. Invented metrics are simultaneously dishonest and — because they are corroborated nowhere — largely useless to the model.

Where this leaves us

It leaves us with an empty case-studies page, which we are aware is an unusual thing for a marketing network to have and an unusual thing to draw attention to.

We have worked with substantial organisations — the client roster is real and it is public. What we do not yet have is a set of documented outcomes that meet the standard described above: baseline, control, mechanism, and the client’s explicit approval to publish. Until we do, the page will say so, plainly, rather than being filled with the kind of confident, unverifiable, generated-looking claims that the rest of this essay argues are worthless.

We would rather be the firm with the empty page and the intact reputation. In a market where every claim is free, that is not modesty. It is the position with the best long-term returns.