AI Discoverability Playbook

Help a website become easier for search-backed AI systems to discover, retrieve for the right question, quote accurately, and verify. Treat classic technical SEO as the foundation, then optimize the passage and evidence layer that answer engines depend on.

Do not promise rankings or citations. Do not treat model-training access, search retrieval, and answer citation as the same mechanism.

Choose the task mode

• Audit: inspect a site, page, rendered output, or codebase and identify the gates that block discovery, retrieval, extraction, or trust.
• Plan: turn audience questions, first-party expertise, and evidence gaps into a topic map and prioritized publishing plan.
• Create or revise: produce a page whose main answer is self-contained, evidence-bounded, and easy to extract without losing qualifiers.
• Implement: improve crawl controls, canonicals, sitemaps, server-rendered content, semantic HTML, metadata, structured data, or internal links.
• Measure: define a repeatable prompt panel and evaluate qualified citations, not isolated screenshots or domain mentions.

For a comprehensive audit, content plan, page rewrite, or measurement design, read references/playbook.md. For a narrow question, load only the relevant section.

Work from evidence

Inspect the artifacts the user supplied before recommending changes. Depending on the request, this may include live pages, raw or rendered HTML, robots.txt, XML sitemaps, HTTP headers, canonicals, structured data, internal links, analytics or search-console exports, and the source that generates the site.