What AI search changes about SEO, and what it doesn't

Google describes its generative AI features as "rooted in our core Search ranking and quality systems." These features retrieve pages from the Search index to ground a response, a method known as retrieval-augmented generation, then show links to supporting pages. They may also use query fan-out, which runs several related searches across subtopics.

To be a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to show a snippet, and the site must not be opted out of generative AI features in Search Console settings. Beyond that, Google lists no special technical requirements. Meeting those requirements doesn't guarantee your page will appear.

Google goes further on labels. Its guide says that optimizing for generative AI search "is optimizing for the search experience, and thus still SEO." We think the name matters less than the work. Call it AEO or SEO, and judge each task by whether it helps people and search systems find and trust your content.

Google's guide has a section on what you don't need to do, and it's useful to keep open when you review a proposal. If someone sells one of these tactics as the way to win on Google, ask what evidence they have.

Structured data isn't an AI trick, and it's still worth doing. Google says it isn't required for generative AI search and that there's no special schema.org markup for it. Google also says to keep using it because it supports eligibility for rich results. Make sure it matches the visible text on the page.

The case for llms.txt depends on who you want to read it. Google says Search doesn't use these files, and that keeping one for other services neither helps nor harms your Search visibility. If you publish one, do it because another tool or agent you care about can use it, and expect no change in Google.

The biggest shift is in the type of page worth creating. Google warns against content that "could easily be produced by a generative AI model," and its example of commodity content is a generic list of tips for first-time homebuyers. An experienced take on why one buyer waived an inspection and looked inside the sewer line is its example of the opposite. If an AI answer can summarize your page without losing anything, your page has given a reader little reason to click.

Try this on your top 20 informational pages. For each one, ask what the page contains that a summary of competing pages couldn't. That could be first-hand testing, original data, a real process, or an expert's opinion. Mark the pages with nothing unique, then improve, merge, or retire them. This is ordinary content strategy, with a higher bar.

Query fan-out points the same way. A single question can spawn related searches, which is a reason to answer the follow-up questions a buyer would have on a strong page. It isn't a reason to publish a thin page for every variation.

You now need to look at AI visibility and business results separately. The Generative AI performance report in Google Search Console shows impressions in AI Overviews and AI Mode, broken down by page, country, date, and device. Traffic from these features also counts toward the Web search type in the main Performance report.

Impressions won't tell you whether AI visibility brought in business. Compare the pages with the most AI impressions against their conversions in your analytics. Google says clicks from pages with AI Overviews are higher quality, and you can test that claim on your own site. We'd report AI referral traffic as its own channel for this reason.

Be careful with third-party visibility scores. Prompt tracking tools can still be useful as a sample of what appears for prompts you choose. Treat them as a directional sample, and use your first-party data for decisions.

All of the guidance above comes from Google and describes Google. Other assistants have their own crawlers, sources, and controls, so ranking well in Google doesn't tell you how you appear elsewhere. Testing the tools your buyers use is the only way to know.

OpenAI separates its crawlers by purpose. OAI-SearchBot surfaces sites in ChatGPT search, while GPTBot gathers content that may train its models, and you can allow one and block the other. A site that opts out of OAI-SearchBot won't appear in ChatGPT search answers, apart from navigational links. So a blanket block on AI bots can hide you from a tool your buyers use.

Microsoft offers its own view too. The AI Performance report in Bing Webmaster Tools shows when your content is cited in Copilot and AI summaries in Bing. It doesn't show clicks, so a citation alone can't tell you whether anyone visited. Each tool's results need to be read separately.

We think the first step is to confirm the basics, then spend your time on content and measurement. This order keeps you from paying for hacks while a technical problem blocks the pages that matter.

1. 1. Confirm your key pages are indexed and can show a snippet, and that nobody has opted your site out of generative AI features in Search Console. Check that robots.txt, your CDN, and your hosting all allow crawling.
2. 2. Review pages for commodity content, and add what only you can say.
3. 3. Keep Google Business Profile and Merchant Center information current where they apply.
4. 4. Set up Search Console reporting, and connect AI traffic to conversions.
5. 5. Decide which non-Google AI crawlers you'll allow, and write the decision down.