AI search readiness (GEO): what is known, what isn't and what to do now

What GEO means

Generative Engine Optimization (GEO) is the practice of making content easy to find, understand and cite for AI-assisted search and answer engines. The term is new; most of the underlying work is not. It overlaps heavily with good search engine optimization (SEO) and clear writing.

What Google says

Google's documentation on AI features states that no additional requirements or special optimizations are needed to appear in AI Overviews or AI Mode, and that sites do not need to create new machine-readable files, AI text files or special markup for these features. The recommended approach is the same fundamental practice as for Search: pages that are indexed and eligible to be shown with snippets, helpful content made for people, crawlable pages with internal links and a good page experience.

Crawler access is a decision, not a default

AI providers run different crawlers for different purposes. OpenAI, for example, documents OAI-SearchBot as the crawler used to show websites in ChatGPT search features, and GPTBot as the crawler used to collect content that may be used to improve its models. Google documents Google-Extended as a separate control over whether crawled content may be used to train Gemini models, and states that it does not affect inclusion or ranking in Google Search.

These controls let you separate two decisions: whether you want to be found in AI-assisted search and whether you want your content used to train models. Review robots.txt with both decisions in mind, and check that a firewall, a security plugin or a CDN is not blocking crawlers you meant to allow.

Content that helps machines and people

• An explicit definition near the top of each main page: "X is..."
• One clear topic per page, with descriptive headings.
• Direct answers to the questions buyers actually ask, followed by the detail.
• Original information: your methodology, your data, your documented experience.
• Transparent claims: sources for facts and clear labels for opinions and hypotheses.
• Structured data that describes the page accurately. Google recommends JSON-LD and states that structured data makes a page eligible for rich results but does not guarantee them.

What remains uncertain

• How each AI system selects and weighs its sources is not fully documented, and it changes often.
• Measurement is still immature: referral traffic data from AI assistants is incomplete and many answers generate no click at all.
• Some sites and tools publish proposed conventions such as llms.txt (a Markdown file that summarizes a site for language models), but Google states that its AI features do not require files of this kind. Treat them as low-cost experiments, not requirements.

We publish an llms.txt file on this site because it costs little to maintain and may help some agents navigate it. We do not claim that it improves visibility.

A practical checklist

1. Confirm that key pages can be indexed: no accidental noindex tags, correct canonical tags and pages included in the sitemap.
2. Review robots.txt separately for search crawlers and for AI crawlers.
3. Include a one-paragraph definition on each main service page.
4. Include frequently asked questions only where they answer real questions.
5. Mark up your organization, your services and your articles with accurate structured data.
6. Publish at least one piece of original, verifiable information each quarter: a method, a data set or a documented case.