Knowledge baseSourcesMetehan Yesilyurt

The RRF Top-n Playbook: How to Get Cited by ChatGPT's Web Mode

Author Metehan Yesilyurt Date 2025-09-01 Model state GPT-5, autumn 2025; date estimated, article undated Open source →

Key findings

A model of result fusion as RRF with derived rules ("2x top-40 or 3x top-90").

Checked against our data

The fusion is not observable; the hit counts are markedly lower for us. The RRF thesis is cited in the field (Peec) but remains a hypothesis.

ClaimStatusEvidence
ChatGPT fuses fan-out results by Reciprocal Rank Fusion, k=60, threshold tau>=0.020unverifiableServer-side. refs.ranking_score_present is false for us; no scores in the object.
ChatGPT retrieves 38-65 sources per answeroutdated14-22 URLs per searching question in the standard mode for us.

Related

rudzki-2026-05-05