{
  "kind": "answer",
  "studySlug": "query-translation",
  "slug": "why-do-models-rewrite-so-heavily-before-they-search",
  "title": "Why do models rewrite so heavily before they search?",
  "description": "Because user prompts are often ambiguous, under-specified, or conversational, while retrieval works better with constrained, evaluative, and category-aware phrasing.",
  "lastUpdated": "2026-01-22",
  "lastTested": "2026-01-22",
  "sourceStudyUrl": "/trakkr-research/query-translation",
  "sourceStudyTitle": "How AI Translates Your Questions",
  "claimIds": [
    "query-translation:complete-rewrite",
    "query-translation:year-injection",
    "query-translation:list-conversion",
    "query-translation:hallucination-rate"
  ],
  "relatedSlugs": [
    "answer:does-ai-prefer-expanded-queries-over-shorter-ones",
    "answer:how-should-you-handle-competitor-language-if-ai-will-insert-it-anyway",
    "fact:the-exact-match-rate-is-only-zero-point-one-seven-percent",
    "tracker:query-rewrite-pattern-tracker"
  ],
  "methodologySummary": "Built from 11,521 captured prompt-to-query pairs observed in OpenAI web search calls, with 100% search-query coverage in the sampled dataset.",
  "limitations": [
    "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."
  ],
  "keywords": [
    "query translation",
    "AI search rewrites",
    "prompt rewriting",
    "AI retrieval behavior",
    "why AI rewrites queries",
    "AI retrieval planning"
  ],
  "schemaHints": {
    "pageType": "Article",
    "includeDataset": true
  },
  "question": "Why do models rewrite so heavily before they search?",
  "directAnswer": "Mostly, because user prompts are often ambiguous, under-specified, or conversational, while retrieval works better with constrained, evaluative, and category-aware phrasing. The rewrite layer is where the model resolves that mismatch.",
  "answerSummary": "The evidence is visible in the 31.85% complete rewrite rate and the strong pattern of added year, list, and entity terms, meaning content must be optimized for the machine generated query rather than the raw user input.",
  "keyFacts": [
    {
      "label": "Complete rewrites",
      "value": "31.85%",
      "detail": "3,670 pairs fell into the complete rewrite bucket.",
      "claimId": "query-translation:complete-rewrite"
    },
    {
      "label": "Year injection rate",
      "value": "25.66%",
      "detail": "2,956 queries injected a year term.",
      "claimId": "query-translation:year-injection"
    },
    {
      "label": "List-format conversion",
      "value": "20.13%",
      "detail": "2,319 rewrites added list framing.",
      "claimId": "query-translation:list-conversion"
    },
    {
      "label": "Brand insertion rate",
      "value": "15.24%",
      "detail": "1,756 queries inserted brand names not in the prompt.",
      "claimId": "query-translation:hallucination-rate"
    }
  ],
  "evidenceTable": [
    {
      "label": "Complete rewrites",
      "value": "31.85%",
      "note": "3,670 pairs fell into the complete rewrite bucket."
    },
    {
      "label": "Year injection rate",
      "value": "25.66%",
      "note": "2,956 queries injected a year term."
    },
    {
      "label": "List-format conversion",
      "value": "20.13%",
      "note": "2,319 rewrites added list framing."
    },
    {
      "label": "Brand insertion rate",
      "value": "15.24%",
      "note": "1,756 queries inserted brand names not in the prompt."
    }
  ],
  "whyItMatters": "Understanding query transformation allows operators to align content with the exact parameters models use during retrieval, such as injected years or brand names, directly impacting visibility in AI search results.",
  "whatToDo": [
    "Write for the query the model is likely to search rather than just the wording a human might type first.",
    "Add current-year, format, and category framing where it genuinely helps the page answer the query.",
    "Ensure competitor and category language is present when the model is likely to inject it during retrieval."
  ],
  "faqs": [
    {
      "question": "How often do models completely rewrite the user prompt?",
      "answer": "Models perform complete rewrites on 31.85% of queries, representing 3,670 pairs in the study."
    },
    {
      "question": "What specific terms are most frequently injected during the rewrite process?",
      "answer": "The year injection rate is 25.66% across 2,956 queries, list-format conversion occurs in 20.13% or 2,319 rewrites, and brand insertion happens in 15.24% or 1,756 queries."
    }
  ]
}