[Largest-scale Query Fan-Out Survey in Japan] AI Searches up to 33 Times for a Single Question — 35,482 Data Points Reveal ChatGPT Performs 1.6x More 'Background Searches' Than Gemini

Queue Inc. (Headquarters: Chuo-ku, Tokyo; Representative: Taichi Taniguchi; URL: https://queue-tech.jp/ ) has announced the results of a large-scale survey analyzing the 'Query Fan-Out' (QFO) behavior of generative AI, based on 35,482 prompts submitted to its free QFO analysis tool for the 'umoren.ai' LLMO/GEO (AI Search Optimization) service between February 5 and May 27, 2026.

The survey revealed that ChatGPT and Gemini automatically generate an average of 4.23 and up to 33 different sub-queries in the background for every single user question to collect information. Furthermore, ChatGPT issues approximately 1.6 times more sub-queries than Gemini, confirming that QFO behaviors fundamentally differ between AI search engines. To our knowledge, this is the first survey in Japan to quantitatively analyze QFO at such a scale.

Six Key Findings

Finding 1: AI searches an average of 4.23 times and up to 33 times behind a single question. When a user poses a single question, AI generates an average of 4.23 different sub-queries in the background for information gathering. In the most intensive case, 33 sub-queries were issued for one question, quantitatively demonstrating that the SEO premise of 'one keyword = one search' has completely collapsed in the AI search era.

Finding 2: ChatGPT executes approximately 1.6x more sub-searches than Gemini. Comparing the average QFO counts, ChatGPT issued 5.29 sub-queries versus Gemini's 3.34, a significant difference of approximately 1.58 times. Even looking at the median, ChatGPT performed 4 while Gemini performed 3, clearly indicating that ChatGPT tends to execute more background searches.

Finding 3: 'High QFO' (7+ sub-queries) is almost exclusively a ChatGPT domain. Classifying the fan-out counts into four tiers—'Low' (1–3), 'Medium' (4–6), 'High' (7–10), and 'Ultra-High' (11+), it was found that 93.5% (2,304 cases) of 'High QFO' (7 or more) cases occurred with ChatGPT, while only 6.5% (158 cases) occurred with Gemini. Furthermore, looking at 'Ultra-High QFO' (11 or more), ChatGPT accounted for 10.9% while Gemini accounted for 0.2%.