Research
Research
Original measurement data from ChatGPT, Claude, Gemini and Perplexity — published openly, with the methodology attached.
What we publish
Every figure in a Gekopilot publication comes from stored measurements: the same question set put to four assistants, with each answer recorded in full. We do not publish estimates, and we do not publish a number we cannot trace back to an answer.
Publications are versioned. When a methodology changes, the version changes with it and earlier editions keep saying what they said.
Publication types
- Sector Index
- A visibility ranking of the brands in one sector across AI assistants. Same question set, four assistants, one table.
- Brand Battle
- How two brands diverge on identical questions, and which assistant favours which.
- Sector Pulse
- How a sector's measurements move over time. Direction, not a single snapshot.
- AI Search Insight
- Findings about assistant behaviour itself: source selection, consistency, disagreement between assistants.
- Flagship Report
- Long-form research with wide scope — several sectors, or a long measurement window.
Latest research
The research pages are currently published in Turkish. The latest edition measures the Turkish sports nutrition market across four AI assistants — and its full report is also available as an English PDF.
Aynı Soru, Dört Farklı Cevap: ChatGPT, Claude, Gemini ve Perplexity Aynı Markaları Önermiyor (Turkish)
Türkiye Sporcu Besinleri AI Görünürlük Raporu — Ekim 2026 (Turkish) · English edition (PDF)
Methodology
Every publication links back to the measurement methodology: how questions are built, how answers become numbers, and what the measurement cannot tell you.