AI visibility score with confidence intervals
Every visibility number in Glossy Data carries its sample size and a 90% interval, and changes are only flagged when they're bigger than normal variation.
See it
Example data in a sample workspace.
ChatGPT names you in about 1 in 3 answers
Mention rate · weekly · 90% interval
Hollow = fewer than 20 answers Engine model change, alerts paused
AI visibility score · Example
Mention rate
Honest Score
We ask 3–5 natural wordings of each question, spread across your topic, on every scheduled run, and pool the answers over a rolling four weeks. Results with few answers are adjusted towards their topic average until they have enough data to stand alone. Older answers count a little less each week, so the score follows real change.
ChatGPT names you in about 1 in 3 answers
Mention rate · weekly · 90% interval
Hollow = fewer than 20 answers Engine model change, alerts paused
“34% · 142 of 412 answers · 90% range 30–38%”. Three labels on every change, and points based on fewer than 20 answers are drawn hollow.
Say you're really named in half of all answers. Pick how many answers we've seen:
At n = 5, the 90% range is 13–87%, a margin of ±37 points.
Normal approximation at a 50% rate, the widest case.
Only when we're at least 90% sure a change is bigger than your threshold (10 points by default), and only after an extra confirmation batch. Every day we also ask each engine a fixed panel of neutral questions (10 per engine): when many unrelated brands move at once, that's the engine changing, so we annotate your chart and hold alerts back.
Alert · significant change
ExamplePerplexity now names you in 41% of answers about “invoicing software for freelancers”.
87 of 212 answers · checked with an extra burst of 16
Monday digest
Details
Labels come from the same list our pricing page and app use. If it says Coming soon, it is not live yet.
| What's measured | Mention rate, top-3 rate and average position, on questions that don't name you |
|---|---|
| Shown with | Sample size (n) and a 90% interval on every number |
| How often | Every scheduled run, pooled over a rolling four weeks |
| Plans | Every paid plan |
| Status | Live |
AI answers change from one ask to the next. Repeated answers give you a more useful picture, while the range shows the uncertainty in the result.
The change is small enough that normal variation could explain it. We show it in grey and don't alert you.
Any point with fewer than 20 answers behind it. We label it and draw it hollow instead of hiding it.
Yes. Questions closer to a buying decision count for more by default, and you can see and edit the weights.
Paste your website. We find your market and category, ask each engine, and show you every answer.
