What AI engines actually read on your site.

THE SHIFT

People can ask buying questions in ChatGPT, Claude, Perplexity or Google’s AI search features. Depending on the question and product, an answer may include sources. GEO — generative engine optimization — describes work on discoverable, readable and well-supported content. The practical foundation remains ordinary SEO: accessible pages that answer real questions and clearly identify their sources.

Start with access and useful content. Use the six dimensions below to find issues worth investigating; confirm the findings against your pages and the relevant provider’s guidance.

THE SIX CHECKS

1. 01

   AI crawler access

   Review robots.txt rules for the crawlers relevant to your goals. Training crawlers and search crawlers have different roles; allowing every AI user agent is not a universal requirement.
2. 02

   llms.txt

   An optional plain-text index for tools that choose to use it. Google does not require llms.txt for its AI search features. Treat it as a publishing convenience, not a citation signal.
3. 03

   Structured data

   Use JSON-LD to describe your organization, articles and authors accurately. Keep it consistent with visible content. No special schema guarantees inclusion in AI answers.
4. 04

   Metadata & crawlability

   Title, description, canonical, Open Graph, sitemap, indexability. The unglamorous basics that decide whether you get read at all.
5. 05

   Answer-ready content

   Clear heading structure, question-style sections, FAQ blocks, direct answers near the top. The shape LLMs pull from when they summarize.
6. 06

   Authority & entity signals

   Named authors, meaningful dates, sources and measured results make claims easier to check. Publish numbers only when you can explain where they came from.