Not sure where to start? Take a guided tour.
AI Site Grader
AI Query Scanner
llms.txt GeneratorMost common query rewrite patterns
Benchmark data indicates that AI query rewrites are dominated by heavy replacement, list-style framing, business-term insertion, and freshness signals. The model rarely preserves the original prompt, instead transforming vague inputs into evaluative, time-bound, or geographically specific retrieval searches.
Research tracker page for the study How AI Translates Your Questions, focusing on the most common query rewrite patterns.
The key numbers.
- Complete rewrites
- 3,670 pairs fell into the complete rewrite bucket.
- 31.85%
- Year injection rate
- 2,956 queries injected a year term.
- 25.66%
- List-format conversion
- 2,319 rewrites added list framing.
- 20.13%
- Best-of conversion
- 2,328 rewrites added best-of framing.
- 20.21%
The ranked view.
- 120.21%
Best-of framing
The model frequently turns vague prompts into evaluative searches.
- 220.13%
List framing
List language is one of the most common retrieval transformations.
- 325.66%
Year injection
Freshness is a recurring rewrite habit.
- 415.24%
Brand insertion
The model often jumps straight to likely market leaders.
- 513.19%
Location addition
Queries are often narrowed geographically before retrieval.
What changed
Exact Match and Complete Rewrite Rates
The exact match rate was recorded at 0.17 percent, with only 20 of 11,521 pairs matching exactly. Conversely, complete rewrites accounted for 31.85 percent of the dataset, representing 3,670 pairs.
Freshness and Brand Insertion Metrics
Year injection occurred in 25.66 percent of queries, totaling 2,956 instances, indicating a recurring habit for freshness. Brand insertion was observed at a rate of 15.24 percent, or 1,756 queries, as the model often targets likely market leaders.
Framing and Location Transformations
Best-of framing reached 20.21 percent, turning vague prompts into evaluative searches. List framing followed closely at 20.13 percent. Additionally, location addition was noted in 13.19 percent of queries, narrowing searches geographically before retrieval.
| Measure | Result |
|---|---|
| Exact match rate Only 20 of 11,521 pairs matched exactly. | 0.17% |
| Complete rewrites 3,670 pairs fell into the complete rewrite bucket. | 31.85% |
| Year injection rate 2,956 queries injected a year term. | 25.66% |
| Brand insertion rate 1,756 queries inserted brand names not in the prompt. | 15.24% |
Population and method
Built from 11,521 captured prompt-to-query pairs observed in OpenAI web search calls, with 100% search-query coverage in the sampled dataset.
Where the evidence stops
This study reflects captured search-query transformations rather than every hidden retrieval step a model may perform.
Some inserted entities reflect category heuristics or market leaders rather than final recommendation outcomes.
The underlying dataset is concentrated in commercial and research prompts rather than all possible conversational traffic.
Cite this tracker.
Source study: Query Translation“Benchmark data indicates that AI query rewrites are dominated by heavy replacement, list-style framing, business-term insertion, and freshness signals. The model rarely preserves the original prompt, instead transforming vague inputs into evaluative, time-bound, or geographically specific retrieval searches.”
Mack Grenfell · Most common query rewrite patterns
How AI Translates Your Questions · Trakkr Research · Updated 22 January 2026