Search queries behind AI answers

An AI platform can search the web using phrases different from the customer question you track. The Search queries behind AI answers block reveals the search phrases exposed in the response data that Searcherries collected.

Find this block in your project's Overview.

Customer questions and AI search queries

A customer question is the input you choose to monitor. A recorded search query is a phrase the AI platform used while looking for information to answer that input.

For example, the tracked question might be:

Which payment services are best for a small online store?

The platform might research phrases such as “small business payment providers” or “online store payment processing fees.” These are illustrative examples, not guaranteed searches for that question.

Read the block

1. Choose a period in Overview.
2. Leave All questions selected, or choose one tracked question.
3. Expand a question in the block.
4. Review its phrases, platform indicators, and repetition counts.
5. Use pagination to inspect more results.

The list shows up to ten search-query rows per page, grouped under their associated customer questions. A question with many recorded queries can therefore appear on more than one page.

Unique queries counts distinct recorded phrases for the question. Recorded searches includes repeated observations. A repetition indicator such as 3× means the phrase was recorded multiple times; it is not a public search-volume estimate.

How the history grows

New project checks can add more queries. Searcherries records phrases when a platform exposes them in the response data. Coverage varies by provider and by answer, and older checks may contain no query history.

“No search queries were captured for this period” means no query records are available for the current scope. It does not prove the AI platform did no research.

Use queries to improve your tracking

Look for terminology, audiences, locations, and subtopics that recur. If the searches consistently focus on the wrong product category, make the tracked question more specific. If a relevant need appears repeatedly, consider adding a separate customer question for it.

Do not replace every question after a single observation. Keep a stable set long enough to compare results across updates.

This block shows platform research activity. To see searches made by people on Google, use Questions likely to surface AI answers in SEO Performance.