An investor has twenty minutes and no way to evaluate your codebase. A procurement lead has a shortlist and a career risk if they pick wrong. An engineer deciding whether to join has your website and a Glassdoor page.
None of them can assess the thing that actually matters. So they assess proxies.
Rory Sutherland puts it plainly: when people can't judge a product's quality, they judge the seller's character. That sounds unfair until you realise it's rational. Character is observable. Quality isn't, not yet.
This chapter is about what you're signalling while people are busy judging.
Costly signals
Sutherland's idea, borrowed from evolutionary biology. A signal is only believable if it would be expensive or difficult to fake.
Anyone can write "enterprise-grade" on a homepage. It costs nothing, so it proves nothing, and experienced buyers discount it automatically.
But thorough API documentation is expensive. A detailed public changelog is expensive. A status page with real history is expensive. A carefully written error message is a signal that somebody cared about a page almost nobody sees.
The pattern: cheap signals are claims, costly signals are artifacts. Buyers have learned to ignore the first and read the second.
Useful way to audit yourself. Take everything on your site meant to build confidence and ask: could a company with a worse product have produced this in an afternoon? If yes, it isn't doing any work.
"Bank-grade security. SOC 2 compliant. Enterprise-ready."
Three claims, zero evidence, on a page with no other detail. Every competitor says this.
And on proof: a wall of customer logos. Tells a buyer people bought. Says nothing about what happened next, and logos are easy to acquire.
A trust page listing your actual subprocessors, your incident history, where data is stored, and who to email about it.
Nobody fakes this, because faking it takes as long as doing it.
And on proof: three case studies with named people, real numbers, and one thing that went wrong. The admission is the signal. It's the part a marketing team would have cut.
Audit your signals this week
Make the list. Everything a stranger encounters while deciding whether you're real: homepage, pricing, docs, trust or security page, changelog, case studies, careers page, your last three months of public writing, the product's free tier or demo.
Score each one cheap or costly. Be honest. Claims are cheap. Artifacts are costly.
Find the missing ones. For SaaS, the usual gaps are a real trust page, clear pricing (or a clear reason there isn't any), and documentation that goes deeper than the happy path. Their absence gets read as a signal too.
Fix the pitch deck for the clock. DocSend's data says investors average three minutes forty-four seconds on a seed deck, and only 58% get viewed to completion. Everything load-bearing goes in the first half, and every slide has to earn the next one.
Then pick one costly signal and overbuild it. Not all of them, you don't have time. One. Usually documentation or one exceptional case study. Depth in one place reads as competence everywhere.
Teardown
Linear
Linear is the clearest example in SaaS of quality as a deliberate signal.
Karri Saarinen has said they started with quality, and then found that people noticed precisely because it's rare for startups. The product's speed, the keyboard shortcuts, the animation timing, none of it was marketing. But all of it was read as evidence.
What's worth copying isn't the aesthetic, it's the logic. Saarinen makes the comparison to packaging: the care is visible before you use the product, and it sets the expectation you then experience it through.
For a founder doing this alone, that's the practical lesson. You probably can't outspend anyone on brand. You can be conspicuously careful with one thing, and let people draw the inference.
What you leave with
What you now have. A credibility audit: every signal you're sending, sorted by whether it's cheap or costly, with the gaps named and one thing chosen to overbuild.
The test. Show your site to someone whose job would be at risk if they chose a bad vendor. Ask where they'd hesitate. Their hesitation is your list.
Where this usually goes wrong. Founders add more claims instead of more evidence. If your response to looking small is stronger adjectives, you'll look smaller. Specificity is the only thing that reads as confidence.
What it takes. A day to audit, then a few weeks of background work to close the gaps. This is the chapter you can do in the margins of other work, and it compounds.
