AEO / GEO · simonwillison.net
Guide for this site
Score 30. AEO cites you. 402 pays you. The report shows how this enters the estimate. Without charging they still read for free.
Already in place: Token efficiency, Parseable robots.txt, sitemap.xml, Clear access policy.
So more AI crawlers come in
Partial · Directives for AI bots
OAI-SearchBot/PerplexityBot (AEO) are not GPTBot (training). Name them.
Allow search bots; Disallow training if that is your policy.
Missing · Content Signals
ai-train / search / ai-input is the policy new crawlers read.
Content-Signal: ai-train=no, search=yes, ai-input=yes (adjust to your case).
Partial · Rules by bot type
Search, agent, and training should not share a single rule.
Separate User-agent groups for search vs training.
So they cite you (AEO / GEO)
Partial · Markdown negotiation
Accept: text/markdown saves tokens versus navigation-heavy HTML.
If Accept includes text/markdown, respond with markdown for the page.
Missing · Quality llms.txt
An llms.txt with H1, summary, and curated links is the site map for agents.
Publish /llms.txt with H1, a blockquote, and 15–25 important URLs.
Missing · Structured data
Organization/Article/Product/Offer is what answer engines cite.
Minimum JSON-LD Organization; add Article or Product by mode.
Missing · AEO / GEO signals
Article, FAQ, VideoObject, or og:title is what answer engines cite. That is not charging.
JSON-LD Article/FAQPage/HowTo/VideoObject and canonical. Disallow training if you do not sell it.
Partial · Link headers (RFC 8288)
Link: rel=describedby / canonical helps discover llms.txt and feeds.
Add Link toward /llms.txt and the canonical.
Prompts for coding agents are in the report or Fix Pack (half price in early access).