Average prompt-to-query similarity is only about 25% | Trakkr Research
Most of the wording changes substantially before retrieval.
Methodology: Built from 11,521 captured prompt-to-query pairs observed in OpenAI web search calls, with 100% search-query coverage in the sampled dataset.
Claim
Average prompt-to-query similarity was 25.24% in the query translation study.
Why it matters
It explains why pages built only around the literal prompt phrasing often underperform in AI search.
Supporting metrics
| Metric | Value | Context |
|---|---|---|
| Average similarity | 25.24% | Average Jaccard similarity between prompt and search query. |
| Prompt-query pairs | 11,521 | Captured prompt-to-query transformations. |
Related pages
Continue through the same study cluster.
- how aggressive are ai query rewrites - Related answer page
- does ai add current year terms to searches - Related answer page
- query transformation severity tracker - Related tracker page
Data & Sources
- How AI Translates Your Questions - Flagship study behind this page
- Page JSON - Machine-readable companion file