How we measure

Last updated: August 24, 2026

Recommended scores come from asking real AI assistants real questions and reading what they say back. This page explains exactly what that number is, and what it is not.

Where the score comes from

For each of your locations we hold five customer questions, the kind someone would actually type when looking for a business like yours. On every check we send those questions to four AI assistants: ChatGPT, Google AI Overviews, Perplexity and Claude. We read each answer and record one thing: whether your business was named in it.

Your score is the share of those answers that named you. Nothing is estimated, modelled or inferred. If the score says 40 percent, it means that in the answers we collected on that check, four in ten named you.

We ask each question up to three times on each assistant, because one answer is not a measurement. What you see is the result across those runs.

AI answers vary, and that is normal

AI answers change constantly — the same question can name you one day and not the next. Judge the trend across checks, not any single check.

This is not a fault in the assistants and it is not a fault in our measurement. These systems generate an answer each time they are asked rather than looking one up, so two identical questions minutes apart can produce different businesses. It is the single most important thing to understand about the number, which is why the app repeats it on every screen that shows one.

The practical consequence: a single check is a sample, not a verdict. A run of checks pointing the same way is worth acting on. One check that moved is usually just the weather.

A score is a measurement, not a ranking

• It is not a position. There is no list of businesses in order, and being named is not the same as being first.
• It is not a guarantee. It describes what happened when we asked, not what will happen when your customer asks.
• It is not a promise of customers. Being named is a good sign. It is not a lead.
• It is not comparable to a Google Maps rank, a search ranking, or anyone else's AI visibility score. Different questions and different methods produce different numbers.

We do not control what any AI assistant says

We have no relationship with, and no influence over, the AI assistants we query. We cannot make one name your business, remove a competitor, or change how any of them decide what to say. Nobody can honestly promise that, and Recommended does not.

What the product does is show you what they are saying now, what changed since last time, and which of the things known to influence these systems you have not done yet. Doing those things improves your odds. It does not guarantee an outcome.

Two details worth stating plainly

• The Google AI Overviews score is measured through Google's Gemini model, the same generator behind the AI Overviews block, asked directly rather than read off a search results page.
• The questions are carried to each assistant by a vendor, DataForSEO, rather than by us calling each provider one at a time. The questions and the answers are the same either way. Our Privacy Policy lists everyone involved.

Questions about any of this

If a number does not match what you see when you ask an assistant yourself, that is expected, and we would still rather hear about it.

Local SEO Bot LLC
832 Callaway Lane, Sunbury, OH 43074
support@recommended.so