What evidence makes company content more credible and verifiable?

Almost every company page claims the same things: quality, experience, attention to the customer. None of these phrases is verifiable, and that's why none of them convinces someone who has to choose a vendor — neither a person comparing three sites, nor a system that has to decide which source to cite. The difference is made by evidence: concrete elements the reader can check. This guide helps you understand what evidence you already have, what you can produce, and which claims are worth removing because you have none for them.

Every claim has a cost: evidence

The method is simple and uncomfortable. You take a page, underline every sentence that claims something, and write next to it what backs it up. The sentences left with nothing next to them fall into three types: those that can be proven with little effort, those that can be rewritten to be more precise and true, and those that need to be removed.

Rewriting is the most rewarding move. "We respond quickly" says nothing; "a technician the same day for requests received by ten a.m., in the provinces where we have a team" says something that can be verified — and also challenged, which is exactly what makes it credible.

Four types of evidence

Not all evidence carries the same weight, and the difference lies in how easy it is for a reader to check it.

• The document: a certification, an authorization, a specification that was met. It can be checked with whoever issued it, and it doesn't depend on your word.
• The measurement: a data point you collected, with the method stated — how many cases, over what period, with what criteria. The method is part of the evidence, not a detail.
• The case: a piece of work told through the starting problem, the constraints, what you did, and how it went. It's the most effective evidence for someone who has to choose, because they recognize their own situation in it.
• The external reference: a citation, a mention, participation in a project, a genuine review. It doesn't depend on you, and that's why it carries more weight.

Below these is material that gets mistaken for evidence and isn't: years in business, a list of logos with no context, certificates of participation, statements about your own values. They aren't false, they simply don't prove you know how to do that work.

The claim-to-evidence matrix

An example of this work applied to a services page. The first three rows are acceptable with a few additions, the others need to be rewritten or removed.

The exercise is done one page at a time and has a useful side effect: it forces you to ask your technical staff how things really stand, and better material than what the site was saying almost always turns up.

The method and its limits

When you publish a data point of your own, the evidence isn't the number: it's how you obtained it. A few lines are enough, and their absence makes even an honest figure look suspect.

• What you measured: how many cases, which ones, over what period.
• How you measured, and what you excluded from the count.
• What the number doesn't say. This is the part that increases credibility instead of reducing it.

The same goes for the cases you tell. If a result depends on particular conditions — a customer whose records were already in order, an industry with little seasonality — saying so protects you and informs the reader. A case presented as replicable anywhere generates expectations someone will have to manage during the deal.

Who signs it

Content with no author asks for trust without offering anything in return. You don't need a biography: you need the elements that make whoever wrote it relevant.

• The name and role of whoever wrote it or of whoever reviewed the technical content.
• The experience relevant to the topic: how long they've worked on it, what kinds of jobs they've handled.
• The date of the last update, and what changed if the content was substantially revised.

In many small companies, the real author is the owner or a technician who doesn't enjoy writing. The practical solution is the interview: someone gathers the material, someone else reviews and signs it. The signature of whoever has the expertise is worth more than that of whoever typed the text.

Your own evidence, citation, promotion

Three different registers that, mixed together on the same page, cancel each other out.

• Your own evidence is what you did and can show. It should be presented with its limits.
• The external citation is what independent sources say. It should be attributed so the source can be traced, and it shouldn't be bent to say more than it actually says.
• Promotional communication is legitimate, but it needs to be recognizable as such. An offer is an offer: it shouldn't be dressed up as a study.

The trickiest case is a citation used as indirect support: a data point taken from an authoritative report and placed next to your own offer to imply something the report doesn't actually say. Anyone who checks will notice, and from that moment the rest of the page becomes suspect too.

What this guide doesn't cover

This guide covers the quality of evidence and how to present it. There's no method that guarantees a content will be cited by an automated answer system, and nothing written here implies buying links or mentions: bought evidence isn't evidence. How to present completed work so it serves sales has its own guide, as does comparing yourself with a competitor who appears in generated answers while you don't.