How does Kendal.io score a business on the AI Visibility Leaderboard?

We ask real AI answer engines the questions a local buyer would ask, then count how often each business is named. A business's score is its share of all the answers we collected, out of 100. So a score of 30 means the business was named in 30% of the AI answers for that market.

Which AI engines does Kendal.io measure?

Up to five of the answer engines people actually use: Anthropic's Claude, Perplexity, OpenAI's ChatGPT, Google's Gemini, and Google's AI Overviews. We measure through their APIs as a consistent, repeatable proxy for the consumer apps.

How often are the leaderboards updated?

Every quarter. Each page shows the date it was last measured and the date the next scan is due. A leaderboard is a point-in-time snapshot of what the AI said on that date.

The method behind the leaderboards

How we score

Q: Can a business pay to change its ranking?

No. We take no payment to move, add or remove any business on a public leaderboard, ever. A business can hire us to genuinely improve its real AI visibility, and we then re-measure honestly, but that never edits the board by hand.

Independent measurement · last reviewed 5 July 2026

Our leaderboards rank businesses by one thing only: how often the AI names them when a real customer asks. Here is exactly how we measure that, so you can check our working.

What we measure

When someone asks an AI assistant for a good business in your field and your town, the assistant answers with a handful of names. We measure who gets named. Nothing else. Not how good the business is, not how much we like it, not who paid. Just: when the AI answers the questions buyers actually ask, whose name comes up?

The questions

For each market (an industry in a city) we write a bank of real buyer questions: the natural-language things a person types or says when they want a recommendation. A mix of the broad ("who are the best [cosmetic dentists] in [Liverpool]?") and the specific (a named treatment, a price question, an "near me" question). We ask the same questions of every engine, so every business is judged on the same test.

The engines

We put those questions to up to five of the AI answer engines people actually use:

Claude (Anthropic)
Perplexity
ChatGPT (OpenAI)
Gemini (Google)
Google AI Overviews (the AI answer at the top of a Google search)

We measure through each engine's official interface as a consistent, repeatable proxy for the consumer apps. AI answers vary a little from one run to the next, so a single lucky mention should never make or break a score. Measuring across five engines and a bank of questions evens that out.

The neutral judge

Reading an AI answer and deciding whether a business was really named is a judgement, not a word-search. "They're near Smileworks" is not the same as being recommended. So every answer is read by a neutral judge model that decides, answer by answer, whether each business was actually named, and never just matches text. We check the judge against control questions on every run to make sure it stays honest.

The score

A business's score is its share of all the answers we collected, out of 100. If we gathered 80 answers for a market and a practice was named in 33 of them, it scores 41. That is the whole calculation. No weighting, no secret sauce.

One thing the scores show again and again: even the market leader usually appears in well under half the answers. That is not a rounding error, it is the point. AI search is new ground, and in most local markets nobody owns it yet. The board is honest about that on purpose.

Kept fresh

Each leaderboard is a snapshot of what the AI said on the day we measured it. AI answers move, so we re-score every market every quarter. Each page shows the date it was last measured and the date the next scan is due. If you are reading an old snapshot, that is why the dates are there.

Our independence pledge

We take no payment to change a public leaderboard. Not to move a business up, not to add one, not to remove one. Ever.

A business can hire us to genuinely improve its real AI visibility, and when we do, the same tool re-measures it honestly, in public, before and after. But the numbers on the board are only ever the numbers the AI gives us. If we ever sold a ranking, the whole thing would be worthless, including to the businesses that pay us.

The honest caveats

A leaderboard measures visibility in AI answers. It is not a review of a business's quality, skill or care, and it is not an endorsement or recommendation from us.
AI answers change over time and vary between runs. We sample across engines and questions to smooth that out, and we date every snapshot.
We measure through official interfaces rather than the exact consumer apps, as a consistent proxy. Where they differ, we would rather under-claim than over-claim.
Being missing from a board is common and fixable. It usually means the AI has not read the right pages yet, not that a business is doing anything wrong.

Something look wrong? Tell us

If your business is named on a board and something is off, or you are missing and think you should not be, email hello@kendal.io. We will check it against the data and fix any genuine error quickly. You can also ask for a fresh, free scan of your own site: where the AI names you, where it names a rival, and the terms worth winning first.