The TRUST framework

Why AI names some brands and leaves out others. Five things it looks for.

We ask AI the questions your buyers ask, then ask why it left you out. Its reasons fall into the same five groups again and again. We call them TRUST.

True facts

One consistent, correct version of who you are, everywhere AI looks.

When it goes wrong: AI gets your location, owner, prices or services wrong, often because an old listing still says so. It can do real harm: in one charity's report, AI gave an out of date helpline number.
How we check it: We ask AI about you directly, then compare each answer with facts you have confirmed.
What fixes it: Correct your own site first, then the listings AI is reading, then keep one version everywhere.

Real proof

Named case studies, clients, results or impact, so AI has evidence, not claims.

When it goes wrong: AI names a rival because it can point to a named client or a result. All it can find for you is what you say about yourself.
How we check it: We read each answer for the proof it shows for you, and for the brands named instead.
What fixes it: Publish named case studies with real figures, on pages AI can read.

Unmistakable focus

Clear about what you do, for whom, and what kind of organisation you are.

When it goes wrong: AI sees you as a generalist, or as a different kind of organisation. So when someone asks for a specialist, it picks someone else.
How we check it: We group the questions into themes to see where you are named and where you are not, then compare how AI describes you with how you describe yourself.
What fixes it: Give each service or sector its own clear page, and say plainly who it is for.

Seen in the right places

On the independent lists, directories and press that AI searches when buyers ask for options.

When it goes wrong: AI searches for "best of" lists and round-ups, and picks from what it finds. One sportswear brand we tested was on 2 of 14 such lists. Its most named rival was on 6.
How we check it: We read each independent list AI found, and check whether you and your rivals are on it. Lists on a rival's own site don't count.
What fixes it: Pitch to the lists that turn up for your buyers' questions, starting with the ones your rivals are on.

Trusted by others

Reviews, awards, accreditations and recommendations published by someone other than you.

When it goes wrong: The only praise AI can find is on your own site, and it gives that little weight.
How we check it: We read each answer for the reviews, awards and accreditations it mentions, for you and for the brands named instead.
What fixes it: Collect reviews on independent sites, enter the awards your buyers respect, and earn recommendations.

Five ratings, no overall score

A single number hides what to do next. Being Strong on facts and Weak on proof calls for different work from the other way round.

Rated by rule

Each letter is rated Strong, Fair or Weak from what the answers show, with the evidence alongside. Where there's too little to go on, we say so rather than guess.

What it doesn't claim

Nobody outside the AI companies knows exactly how they choose, and answers change from one go to the next. TRUST is built from what AI says and shows, not from inside knowledge.

See where AI leaves you out, and what to fix first.

We're adding a TRUST scorecard to our reports. Start with a free report for your website, in about three minutes.

Get a free report See a sample report