Generative AI Search Is Still SEO

Netsujo Inc. · Published July 7, 2026 · Updated July 7, 2026

With the arrival of AI Overviews and AI Mode, more people are asking whether generative AI search needs its own, separate optimization — whether to add special schema, or set up an llms.txt file. Reading Google Search Central's repeated guidance closely, the answer is fairly simple: optimizing for generative AI search is an extension of ordinary SEO, not a separate discipline. Google states that no special optimization is required; good content, technical health, and E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) remain the foundation. As of a June 15, 2026 documentation update, Google states that llms.txt is not used by Google Search, and FAQ rich results were discontinued from Google Search on May 7, 2026. Structured data helps machine readability, but Google has not stated that it directly raises AI visibility or ranking. Appearing in AI Overviews cannot be controlled or guaranteed by any specific action.

Google's official position

Google states officially that no special optimization is required for generative AI search. There is no need to build a separate mechanism specifically for AI Overviews or AI Mode; good content, technical health, and E-E-A-T — the same things that have always mattered for SEO — remain the foundation. In other words, there is no independent 'AI search readiness' initiative; steadily building up ordinary SEO is what carries over into generative AI search as well. The starting point is thoroughness on the basics, not a novel trick.

How AI Overviews work

AI Overviews is a mechanism where Google retrieves and synthesizes information from multiple pages in response to a search query and generates an answer, rather than pulling the content of a single page as-is; the answer is assembled across multiple sources. What matters here is that whether a page appears cannot be controlled or guaranteed — the claim that 'doing X guarantees inclusion in AI Overviews' does not hold.

llms.txt is not used by Google Search

In a documentation update on June 15, 2026, Google stated explicitly that llms.txt files are not used by Google Search. It's worth distinguishing what Google is and isn't saying: Google states that there is no need to create one and no special requirement — not, as a blanket claim, that llms.txt has no effect in any context. Its use in other contexts or other services is not addressed by this statement. Limited specifically to Google Search, though, the motivation to prepare an llms.txt file is thin.

Consistency of facts matters more than AI-specific writing

Before adding AI-specific schema or AI-specific writing styles, the more important foundation to secure is consistency of facts and ordinary SEO. If information varies across a site — figures or wording that differ from page to page — it lowers trustworthiness for machines and human readers alike. Keeping facts accurate and consistent is the most solid preparation for generative AI search.

Where structured data fits

Structured data (schema.org) is effective for improving the machine readability of content. However, Google has not officially stated a direct causal link between adding structured data and an increase in AI visibility or ranking — its effect should not be overestimated. Note also that FAQ rich results stopped displaying in Google Search as of May 7, 2026; FAQPage JSON-LD no longer carries any display benefit within Google Search itself. Retaining it may still make sense for machine readability or reuse by other systems, but there is no need to add structured data purely for display purposes.

E-E-A-T and technical SEO as the foundation

For generative AI search too, the support comes from thorough E-E-A-T and technical SEO: primary, first-hand content backed by experience, expertise, authoritativeness, and trustworthiness; a site structure that doesn't obstruct crawling and indexing; sensible internal linking; and page speed. Steadily strengthening these basics tends to be more effective in the end than searching for a shortcut.

Effects on generative AI search should be treated as reference data, not a guarantee. Appearance in AI Overviews can't be controlled, and claims of guaranteed ranking or AI inclusion aren't accurate — measure continuously via GA4 and Search Console and adjust priorities based on the evidence.

AI Overviews: how it works and how to approach it（日本語）

What is llms.txt, and how do you write one?（日本語）

Why technical strength doesn't come across on the web: information fragmentation（日本語）

Measuring AI visibility: why a single snapshot isn't enough（日本語）

Our fact-only SEO/AIO practice log（日本語）