ChatGPT, Gemini and AI Mode disagree on the "best" tool in half of software categories

Описание и применение

Три ИИ‑движка (ChatGPT, Gemini, AI Mode) по-разному определяют лидеров в 10 категориях программного обеспечения, показывая, что рекомендации зависят от источника.

Задача

Different AI answer engines produce conflicting top‑brand recommendations for software categories, making it hard to determine a single best tool.

Результат

In 5 of 10 categories the three engines named different leaders; in the other 5 they agreed but the full rankings diverged. The study shows AI visibility is engine‑, category‑, and country‑specific.

Рабочий процесс

• 1. Prepare a set of 182 buyer‑intent prompts covering 10 software categories.
• 2. For each prompt, send one API call to each engine (ChatGPT, Gemini, AI Mode).
• 3. Collect JSON responses, extracting cited source URLs to distinguish citations from passing mentions.
• 4. Count brand mentions per engine per category.
• 5. Compile per‑category leader tables and full ranked lists.
• 6. Publish results on the public leaderboard and provide methodology documentation.

Доказательства

in 5 of 10 categories, the three engines name a different brand as the leader.

Источник

The pipeline is boring on purpose: prompt set → one API call per engine per prompt → count mentions.

Источник