How Finded measures AI Visibility

A 0–100 score combining mention frequency, intent coverage, model consistency, and website signals, one number to track your visibility over time.

The share of tested AI responses that mentioned your business. A rate of 42% means you appeared in 42% of relevant answers.

Which AI assistants (ChatGPT, Claude, Gemini, Perplexity) recognise and recommend you. Consistency across models matters more than strength in just one.

How you compare to the businesses AI recommended instead of yours, for the same customer intent. This comparison set changes depending on what was asked.

How many distinct customer questions and situations your business is visible for, not just keywords. For an accounting practice that might be a freelancer’s VAT return or a first annual account; for a plumber, an emergency callout or a boiler replacement.

The customer intents where AI consistently recommends your business, worth protecting.

Customer intents where AI rarely recommends your business, the highest-impact places to improve.

Compares your latest scan with previous ones to show whether visibility is moving up or down.

An estimate of how much a recommendation could help, drawn from measured patterns and always shown with its evidence. It is not a guarantee for your case.

An estimate of how much work a recommendation requires, used alongside Expected Impact to prioritise.

How consistently a recommendation's pattern has been observed across businesses, high confidence means well-established, lower means emerging.

No business is recommended for every possible customer intent, by every AI model, every time, a certain amount of variation is normal, even for well-established businesses. A very high score is achievable, but 100 would mean perfect, universal recommendation, which isn't a realistic or meaningful target. Finded focuses on measurable progress, not a perfect number.

The businesses AI recommends instead of yours depend entirely on what's being asked. A different customer intent surfaces a different set of alternatives, so your competitive set shifts from one evaluation to the next. This is expected, it's exactly why we track competitive position per intent, rather than as a single fixed ranking.

Yes. AI models are updated regularly, and their responses can shift as a result, independent of anything you've changed on your end. This is why Finded monitors visibility continuously through repeat scans, rather than treating a single measurement as final.

No. Finded measures and reports what AI models currently return, and provides evidence-based recommendations to improve your visibility. We don't control AI models and we don't attempt to manipulate their outputs, we can't guarantee specific outcomes, only give you an accurate, evidence-based view of where you stand and what tends to help.

Monthly is enough to track meaningful change for most businesses. If you've just implemented recommendations, a rescan lets you verify whether they had a measurable effect, rather than assuming they worked.

A single question gives you one data point, from one model, at one moment. Finded evaluates a structured set of customer intents across multiple AI models on a recurring basis, so you get a measured, repeatable picture, including trend and competitor data, instead of a one-off answer.