Query rewriting · Query reformulation
Query rewriting is the engine’s rephrasing of a user query before retrieval. It is the name used for ChatGPT Search for a mechanism close to query fan-out: the prompt is rewritten into one or more optimized search queries.
Standalone definition · 38 words
How does query rewriting work?
The system transforms the prompt, often conversational and implicit, into queries suited to a search engine: adding missing context, resolving pronouns, specifying the period or place, or breaking it into several searches. According to OpenAI’s description relayed by trade press, ChatGPT search may rewrite the prompt and, if memory is on, use memory items.
Query rewriting or query fan-out?
The two terms describe neighboring mechanisms. Query rewriting emphasizes the transformation of the query, query fan-out the multiplication into parallel sub-queries. In engines, the two can be combined. The names come from providers or practitioners, not a standardized vocabulary.
Consequences for content
Content that is not retrieved by the user’s prompt may be retrieved by the rewritten query. Covering close phrasings and naming entities and context (place, period, audience) increases the chance of being retrieved. You optimize an intent, not a character string.
Frequently asked questions
Can I see the rewritten query?
Sometimes, depending on the interface, some engines show the searches performed. It is not guaranteed.
Which engine uses query rewriting?
All engines with web search do so in one form or another.
Does query rewriting depend on memory?
According to OpenAI, memory when enabled can play a role in rewriting.
Related terms
Further reading
- Checklist GEO and LLM / AI 2026
- 2026 GEO Tools Comparison: The Decision Guide
- ← Back to the GEO glossary (114 terms)