[Independent Study] SEO Managers Focused Only on Search Volume Risk Being Left Behind by AI

EXIDEA (Headquarters: Chuo-ku, Tokyo; CEO: Takuma Ogawa) conducted a survey of 17,693 queries to understand how Google's AI Overview (AIO) changes brand presentation based on 'how the question is asked,' even for the same non-branded search queries (where users do not include a brand name).

In this release, we share four facts that can serve as a reference for SEO and content managers when designing keywords in the AIO era.

In the field of SEO and content management, as teams consider their response to AIO, the following concerns are frequently heard regarding the exposure of their content to AI: - "Should we prioritize measures based on keywords with high search volume, or should we look at queries that are more easily adopted by AI from a different axis?" - "How much does brand appearance in AIO change depending on the query format, such as 'What is...?', 'Recommended', or 'Comparison'?" - "Even for keywords with similar search intent, does the AI's answer content or brand mention differ depending on how the question is asked?"

Key takeaways from the survey results: - In AIO, even for the same non-branded searches, the volume of brand appearances changes significantly based on 'how it is asked.' - Particularly in 'search to choose' types like comparison and list formats, the AI showed a strong tendency to list multiple brands. - In future SEO, it is important to design keyword and article roles not just by search volume, but by looking at 'whether it is a way of asking that makes it easy for AI to recall the brand.'

Survey Overview: - Analysis Blocks: 288,177 (paragraph units within AIO answers) - Dependent Variable: Binary judgment of whether brand/company/product/service names appeared in each AIO block. - Query Classification: 5-axis classification (search intent/query format/industry/brand presence/YMYL judgment) using LLM (gpt-5-nano). - Brand Extraction: Automatic extraction by LLM. - Statistical Test: Query format x Brand mention (Chi-square test, χ²=14,409.1, df=5, p<0.001); Search intent x Brand mention (Chi-square test, χ²=12,721.8, df=3, p<0.001). - Data Acquisition Period: Feb-Mar 2026. - Conducted by: EXIDEA.

1. Prioritize 'Comparison' over 'Definition' to target brand exposure: Even with the same non-branded queries, brand appearance in AIO answers changed significantly depending on the format of the user's question. The most significant difference is between 'Comparison' and 'Definition' types.

AIO Brand Mention Rate by Query Format: - Comparison (e.g., 'A vs B', 'A difference'): 75.9% (6.6 avg unique brands) - List (e.g., 'A recommended', 'A ranking'): 61.0% (5.7 avg unique brands) - Method (e.g., 'A how to', 'A steps'): 56.4% (4.1 avg unique brands) - FAQ (e.g., 'A why', 'A when'): 52.0% (3.3 avg unique brands) - Definition (e.g., 'What is A?', 'A meaning'): 39.3% (2.7 avg unique brands)

Between Comparison and Definition, there was a 36.6pt gap in brand mention rate and about a 2.4x difference in average unique brands.