# Query Fanouts

Inspect the background search queries captured while AI engines answered tracked prompts.

Canonical: https://docs.equalai.io/analytics/query-fanouts
Markdown: https://docs.equalai.io/analytics/query-fanouts.md

Open **Reports → Query Fanouts** to inspect captured background searches. A fanout is a query an AI engine used while answering a tracked prompt. It is not necessarily the original prompt, a user search-volume estimate, or a complete retrieval log.

## Find a query

1. Set the date range.
2. Use **All filters** for **Tags**, **Models**, or **Topics**.
3. Search for text in the query, original prompt, topic, model, or tags.
4. In **All queries**, expand the topic and prompt groups to inspect captured queries and the models associated with them.

Repeated normalized query text is grouped together. Occurrence counts and distinct-query counts answer different questions: a query repeated across runs contributes several occurrences but may remain one distinct query.

## Read the summaries

**Brands** counts fanout rows containing your brand or tracked competitor aliases. This is text matching within captured search queries, not evidence that the resulting answer mentioned or recommended the brand. Short or ambiguous aliases can produce matches that need review.

**Common phrases** highlights frequent words and adjacent two-word phrases from the captured query text, after normalization and removal of common filler. These are not independently measured search demand, user intent, or keyword difficulty scores.

Use a recurring phrase as an investigation lead: open the original [response](https://docs.equalai.io/analytics/responses), inspect its cited sources, then decide whether your tracked questions cover that information need.

## Export

**Export** downloads the filtered rows with topic, original prompt, query fanout, model, provider, surface, position, and capture date. The position is captured query metadata, not a guaranteed organic search rank.

## Troubleshooting

- **No fanouts despite having responses:** some providers do not supply background-query evidence. Absence of captured fanouts does not prove no search occurred.
- **Counts fall after filtering:** the summaries are recalculated from matching rows.
- **A brand match seems unrelated:** inspect the actual query and configured aliases.
- **Common phrases look generic:** review the underlying queries rather than treating the phrase summary as a content recommendation.

## Related guides

[Responses](https://docs.equalai.io/analytics/responses) · [Citations](https://docs.equalai.io/analytics/citations) · [Prompt Explorer](https://docs.equalai.io/ai-visibility/prompt-explorer)