AI visibility margin of error calculator
Free tool
AI engines give a different answer each time you ask. Enter how many answers you collected and how many mentioned your brand, and see the range your true share is likely to sit in.
60% give or take 26 points
6 of 10 answers mentioned you. With 95% confidence, your true share is between 31% and 83%.
That range is too wide to tell a real change from chance. Ask more times before reporting it.
How many answers do you need?
The margin shrinks with the square root of the number of answers, so halving it takes four times as many. These are the answers needed per question, per engine, each time you check.
| Margin | Answers needed (worst case) | At your 60% share | Cost per check |
|---|---|---|---|
| ± 25 points | 16 | 15 | 2¢ |
| ± 20 points | 25 | 24 | 3¢ |
| ± 15 points | 43 | 41 | 6¢ |
| ± 10 points | 97 | 93 | 12¢ |
| ± 5 points | 385 | 369 | 49¢ |
Worst case assumes a 50% share, where answers vary most. Costs use $0.0012 an answer from DataForSEO's queue for ChatGPT, Gemini or AI Overviews, plus $0.00008 for TypeSafe's Jev to read it.
How it works
Each answer either mentions you or doesn't, like a coin that lands one way or the other. The calculator uses the Wilson score interval, which stays honest with small samples and with shares near 0% or 100%, where the textbook formula breaks down.
It assumes each answer is independent. Answers collected in one burst can be more alike than answers spread across days, so treat the range as the least uncertainty you have, not the most.