Trust architecture

The problem it names

In categories where the cost of being wrong is high, consultancy, healthcare, capital, industrial supply, buyers are not primarily evaluating quality. They cannot. They lack the expertise to assess the work before it is done, which is the reason they are buying it.

So they evaluate proxies. How the firm behaves in the sales process, what other people say, whether the published material demonstrates thinking or asserts it, whether anyone will put a name to a claim.

This is not irrationality. Economists have studied it for fifty years. George Akerlof’s 1970 paper on markets with asymmetric information showed what happens when buyers cannot distinguish quality before purchase: they discount everything, and good sellers are driven toward the average [1]. Michael Spence showed the counter-move, which is signalling: a seller demonstrates quality by doing something that would be too costly to fake [2].

Trust architecture is the practical version of Spence’s point. The question is not what do we claim, it is what have we done that would be expensive for an incompetent firm to copy.

The distinction that does the work

A claim is cheap. Anyone can write that they are rigorous.

A signal is costly to fake. Publishing the method so a competitor can check it. Stating what you got wrong. Showing a worked example with numbers in it. Naming the conditions under which your approach fails.

Most marketing in credibility-dependent categories is entirely claims, which is why it does not work and why everyone in those categories sounds identical. If every firm asserts rigour, the assertion carries no information.

The four layers

This is our framework rather than a finding, and it is useful mainly as a checklist.

What you publish. Does it demonstrate thinking or assert competence. A reader should be able to disagree with something specific.

What you admit. A case study with no scar in it reads as an advertisement and gets priced as one. The willingness to document a limitation is itself the signal, because a weak firm cannot afford it.

What others say. Independent corroboration, which you do not control and which therefore carries more weight than anything you publish.

What happens on contact. How quickly you answer, whether the proposal reflects what was discussed, whether you tell them something inconvenient. This is the layer that most often contradicts the other three.

Where it goes wrong

Firms build layer one and stop. A content programme that asserts expertise, no admissions, no independent corroboration, and a sales process that contradicts the published voice.

They treat admission as risk. It is the cheapest signal available and the most under-used, because it feels like weakness in the room where it is approved.

The layers contradict each other. Published material about partnership, proposals written in boilerplate. Buyers notice the contradiction and resolve it in favour of the behaviour, not the publishing.

What to do with it

1. Audit your last five pieces of published material for one falsifiable statement each. If none contains anything a reasonable person could dispute, you are producing claims.
2. Put a documented limitation in every case study. Not a humblebrag, a real constraint and what it cost.
3. Count your independent sources. If every statement about your competence originates with you, layer three does not exist.
4. Mystery-shop your own enquiry process. The gap between layers one and four is usually the whole problem.
5. Ask of every planned asset: would this be expensive for a firm that could not do the work to produce? If not, it is a claim.

If you run the company rather than the marketing

The one question worth taking from this: what do we publish that a competitor who cannot do this work would find expensive to copy?

For most firms in most professional categories the honest answer is nothing. The website, the credentials, the client logos and the language about partnership are all available to anybody, at no cost, regardless of whether they are any good. Which is precisely why buyers discount them.

The things that are expensive to fake are the ones firms avoid: publishing a method in enough detail to be checked, documenting a project that went wrong and what it cost, putting a number on a result that could be audited, saying plainly which clients you are not right for.

Each of those feels like giving something away or exposing a flaw. That feeling is the point. A signal only works because it costs something, and if it is comfortable to publish it is probably carrying no information.

The cheapest place to start is the last one. Write down who you are not for.