AI visibility tools
An AI visibility tool that reads once is reporting a sample. We measured pools of 8 to 53 companies behind a single question.
Repeat one question and count the distinct companies that accumulate. Depending on what a customer types, the pool an assistant draws from runs from about 8 to about 53 - which sets the odds of any one appearance between 11% and 53%.
8–53
companies an assistant can name for one question, depending on which question is asked
What we measured
- Repeating the same question and counting distinct companies, the pool an AI assistant draws from ranges from about 8 to about 53 depending on which question is asked.
- The implied chance that any one company appears in a given answer ranged from 11% on the broadest question to 53% on the narrowest.
- 4 of the 5 pools we measured were still growing at the fourth reading, so every pool estimate is a floor rather than a ceiling.
- With a pool of about 28 companies and 5 names in an answer, a company squarely inside that pool is absent from roughly 4 answers in 5.
What a tool is actually sampling
| The question asked | Readings | Distinct companies after each reading | Estimated pool | Chance of appearing |
|---|---|---|---|---|
| A one-off report | 4 | 5 → 9 → 14 → 18 | ~28 | 19% |
| Monthly monitoring | 4 | 5 → 9 → 10 → 15 | ~12 | 42% |
| Fix my site | 4 | 4 → 4 → 6 → 8 | ~14 | 25% |
| Build me a new site | 4 | 5 → 7 → 9 → 9 | ~8 | 53% |
| Who helps businesses like mine | 3 | 5 → 10 → 18 | ~53 | 11% |
Capture-recapture on repeated readings of one identical question, one trade, one assistant. The middle column is the running union - how many different companies had been seen at all after each reading. 4 of the 5 were still climbing at the last reading, so these are lower bounds.
What this means when you are comparing tools
If an assistant can name about 28 companies for a question and its answer holds 5 names, then a company genuinely in that pool still fails to appear roughly four times in five. A tool that checks once a month and shows you a green tick or a red cross is reporting that coin flip as a state.
The number a tool should be reporting is a rate: how often you appeared across many readings, with the readings visible. That is harder to draw as a dashboard gauge and it is the only version that survives being checked.
Two more things worth asking a vendor. Does it record which model answered - the tier served to one account drifted under us mid-run, without being asked. And does it keep surfaces separate, or pool an anonymous session with a logged-in one into a single score?
Who an assistant currently recommends here
We looked at who is cited on the 5 questions this page answers, across the 7 answers captured for them. The two most frequent citations are a search-tool vendor and a video platform, each appearing in 6 of the 7; an automation directory follows in 4 of 7.
So on tool questions the durable positions are held by a general SEO suite, a video platform and a directory - not by the specialist tools being asked about. The specialists rotate underneath them.
The questions this page answers
ai visibility tools
1,300/moai visibility tool
1,000/movisibility tracking
880/movisibility tools
480/mobest geo tools
210/mo
5 questions, 3,870 searches a month between them. All 7 answers we examined on them were produced by a web search, which means a page can reach every one.
Who an assistant cites on these questions today
- semrush.com 6
- youtube.com 6
- zapier.com 4
- sedestral.com 3
- surferseo.com 3
- wpengine.com 3
- seranking.com 2
- reddit.com 2
- tryprofound.com 2
- addlly.ai 1
26 distinct domains are cited across these questions; the 10 above are the most frequent, with the number of answers each appears in. Counted from the answers captured in our corpus, not from a market census.
What to ask before you buy
- Ask for the raw answers, not only a score. A score you cannot audit is a number someone chose.
- Ask how many readings sit behind each figure, and over what span. One reading is a sample; our repeats of one identical question overlapped 15% on average.
- Ask whether the model tier and the account type are recorded per reading. If they are not, the tool is reporting an uncontrolled variable.
Honest limits of what we measured
- Pool estimates assume every company is equally likely to be caught on any reading, which is certainly false and biases them downward. Read them as lower bounds.
- 3 or 4 readings per question, one trade, one assistant. These are the right order of magnitude, not precise counts.
- We have not audited any competing tool's internals. Everything here is about what the underlying surface does, which is what any tool has to sample.
Questions we get asked
So are these tools useless?
No. The surface is measurable and worth measuring. The caution is specific: a single reading is a sample from a pool of 8 to 53, so any product built on one check per period is reporting noise as a state.
How many readings are enough?
More than we have. 4 of our 5 pools were still producing new names at the fourth reading, so we cannot yet tell you where the curve flattens - and neither can anyone who has not run the same test.
What do you provide instead?
Repeated readings with the raw answers attached, the model tier recorded per reading, and a rate rather than a verdict. Where a number is uncertain we print the uncertainty.
Can we resell this to our own clients?
Yes - the measurement is the same whether it is run for one business or many, and the report can carry your branding. Terms for multiple clients are a conversation rather than a price on a page.
Start with one free reading
One business, real questions, the raw answers back - before you decide on anything paid. If you are running this for a book of clients, say so and we will talk terms rather than quote a page price.
Free reading
Takes about a minute.
Related
- AEO tools and trackers
- LLM and AI search visibility tools
- AI visibility report
- AI search optimization
- All six measurements →
Pricing is per category, because the work differs by how competitive the answers are. See every category we measure and what a reading costs →
Published 31 August 2026. Every figure on this page is recomputed from raw observation files by a single script, so any of them can be traced back to the readings behind it. We make your site citable. We do not promise placement in any assistant’s answer - we measure whether you are in it, and show you the raw answers either way.