All posts Measurement

Why AI answers change from one run to the next

Ask the same engine the same question twice and the sources can differ. That's not a bug — it's why a single check is a sample, not a verdict.

On this page

Run the same question through the same engine twice and you may get a different set of sources. This surprises people who expect a search engine’s steady results. It’s worth understanding, because it changes how you should read any single check.

Answers are generated, not looked up

An AI answer is composed fresh each time. The engine decides whether to search, what to search for, and which of the pages it finds to lean on. Small differences in any of those steps lead to different citations.

What this means for a check

One run tells you what happened that once. It’s a sample, not a final score. A site that wasn’t cited today might be tomorrow, and the other way around. This is exactly why we show “one run, answers vary” on every report, and why we never turn a result into a guaranteed ranking.

Tracking beats a snapshot

Because any one answer is noisy, the signal lives in the pattern over time. If you’re cited in four of the last six weekly runs, that’s meaningful. If you appear once and never again, that’s noise. Weekly tracking turns a series of samples into something you can actually trust.

Run a free check on your site No signup. About 30 seconds. One private report.
Run a free check