Research
Seven measurements nobody else has run.
We sell a reading of how AI assistants describe a business, so we run that instrument on the question of whether such a reading is worth anything at all. Everything below is our own data, published in full, with its method and its limits. Every figure is recomputed from raw observation files by a single script, so any of them can be traced back to the readings behind it.
- 0%
We asked an AI assistant the same question twice, 10 minutes apart. The two answers had zero companies in common.
Overlap between two readings of the same question, 10 minutes apart, on the same model tier
- 8–53
How many companies can an AI assistant name at all? Between 8 and 53, depending on which question you ask.
Estimated companies an assistant can name, depending on the question asked
- 1
One of the companies an AI assistant named has a single referring domain. Off-page authority is not the gate.
Referring domain on the weakest company an assistant named, against 2,095 on the strongest
- 5
Five things we tested that do not predict whether an AI assistant names you.
Explanations tested and ruled out as predictors of being named
- 19.4×
People ask about AI visibility on Google, not inside ChatGPT - by a factor of 19.4 against 2.7 for ordinary business language.
How much more often category vocabulary appears on Google's AI answers than in ChatGPT - against 2.7× for everyday business phrases
- 8.7
An AI answer to a business owner cites 8.7 sources - and most of the seats are held by platforms and forums, not vendors.
Sources cited, on average, in an AI answer to a business owner's question
- 0 of 22
For some trades the assistant reads the businesses’ own websites. For others it never opens them.
Citations that pointed at a business’s own website, in the trade where the assistant reads them least (assisted living communities)
What do we do with them?
These measurements come out of the work we sell. Each page below answers one question people already ask about this category, and carries the number we measured for it.
Why do we publish these?
Five separate explanations for who gets named in an AI answer failed our tests: page structure, sentence-level text matching, a page per trade, the size and shape of the site, and off-page authority. What survived is that assistants cite whoever owns the specific fact the answer needs. These pages are us owning ours.
If you are named on one of these pages and believe a reading is wrong, tell us. We will run the measurement again and publish what it returns, whichever way it goes.
Every category we measure has its own page, with the questions customers put to an assistant there. See all industries we measure →
Run the same instrument on your own business
Every number on these pages came out of the reading we sell. We will run it for your business and show you the raw answers, whichever way they come back.