Sample size (n)
Last updated [DATE]
n is the number of usable answers a figure is based on: answers with status "ok". Every rate in Aivting is shown with its n, because 1 of 3 answers and 10 of 30 are both 33%, but the second tells you much more.
How to: Why numbers differ from ChatGPT
What counts toward n
Only answers with status ok, for active prompts, inside your filters. These are not counted in n:
- no answer: the engine returned nothing usable.
- failed: collection did not complete.
- blocked: not collected because a spend limit was reached.
They are counted separately and never as "not mentioned". If n is 0, no rate is shown.
Example
Brindlepath (fictional) has 10 prompts and one run collects 3 answers per prompt from ChatGPT: 30 answers. 27 are ok, 2 are no answer and 1 failed. n is 27. If 9 name Brindlepath, visibility is 9 of 27, or 33%.
Minimums
Some figures need more answers before Aivting shows them.
| Figure | Minimum |
|---|---|
| Visibility, share of voice | Any n above 0, but read small n with care |
| Average position | 3 answers that name your brand |
| Your site as a source | 10 answers that cite your site and 10 that do not |
| Sentiment | n is the answers that name your brand and have a tone label, not all answers |
| A change between two periods | At least 10 answers behind the figure in each period. See Filters and date ranges |
Reading small n
A rate from a handful of answers moves a lot between runs. Prefer weekly trends over single days, widen the date range, or add prompts. A rate from 3 answers is a hint; a rate from 60 is a measurement of that sample.
Related
Visibility · Runs and answers · Filters and date ranges
ChatGPT is a trademark of OpenAI. Google and AI Overviews are trademarks of Google LLC. Aivting is an independent product and is not affiliated with or endorsed by OpenAI or Google.