Score the page generative engines actually read.
Generative engines like ChatGPT, Perplexity, Gemini, and Claude lift answers from sites they can crawl, understand, and trust. Rook Ranker scores generative engine readiness across 441 live signals — around 163 score a typical page, more as you add locales and bot directives.
Crawl, comprehend, cite.
A generative engine has to find your page, parse meaning from it, then choose to cite you. Each layer is a separate score.
Crawl access
Robots.txt checked individually for GPTBot, ChatGPT-User, OAI-SearchBot, PerplexityBot, ClaudeBot, anthropic-ai, Google-Extended, Bingbot, YouBot, DuckDuckBot.
Comprehension
Schema density, entity reinforcement, llms.txt, semantic chunking, definition led intros. The signals engines use to understand the page.
Citation patterns
TL;DR blocks, question shaped headings, lists, tables, dates, citations, sameAs. The structures that get lifted into answers verbatim.
Ten generative engines, each scored on its own.
Each generative engine respects different bot tokens and crawl rules. We test for each one separately so you see exactly who can read you and who cannot.
- GPTBot, ChatGPT-User, OAI-SearchBot for OpenAI
- PerplexityBot for Perplexity
- ClaudeBot, anthropic-ai for Anthropic
- Google-Extended for Gemini
- Bingbot for Microsoft Copilot
- YouBot, DuckDuckBot, and growing
llms.txt
A curated map of the site for generative engines. Rook Ranker checks if you have one and grades the structure.
Freshness
datePublished, dateModified, visible "last updated" copy. Engines weight recent content heavily.
Knowledge graph hooks
sameAs to Wikipedia, Wikidata, Crunchbase, LinkedIn. The links that bind the entity to authority.
Ingestion chunking
Heading every ~250 words. Short paragraphs. Concise sentences. Engines lift focused chunks.
Citation bait
TL;DR, question headings, definition lists, statistics. Patterns engines quote verbatim.
Entity reinforcement
Brand name in title and body. Org schema with full identity fields.
Allow the right bots.
Robots.txt is checked per generative engine. Block anyone you do not want training. Allow anyone you want citing you.
Ship a clean llms.txt.
A short curated map, written in markdown, with links to the pages you want cited. Rook Ranker scores the structure.
Write the way engines quote.
A definition led first sentence. A TL;DR. Lists for steps. Tables for comparisons. Question shaped subheads.
Tie identity to authority.
Organization schema with sameAs to Wikipedia, Wikidata, LinkedIn. The hook engines need to bind you to a known entity.
Each bot is checked on its own.
Blocking GPTBot but allowing PerplexityBot is a real choice. We surface it, never average it away.
Citation friendly structure earns points.
Lists, tables, definition lists, dates, statistics. Quotable structures.
Read off the rendered HTML.
No simulators. The same page the engines see is the page we score.
Frequently asked.
What does GEO stand for?+
Generative Engine Optimization. The discipline of preparing a page so that ChatGPT, Perplexity, Gemini, Claude, Grok, and similar engines can read it, understand the entities on it, and cite it accurately.
How is GEO different from SEO and AEO?+
SEO is the classics for ranking on Google. AEO is being cited by answer engines. GEO sits next to AEO and focuses on how ready your raw signals are for generative ingestion. Crawl access for AI bots, llms.txt, freshness, knowledge graph hooks, and citation friendly structure.
Which generative engine bots do you check?+
GPTBot, ChatGPT-User, OAI-SearchBot, PerplexityBot, ClaudeBot, anthropic-ai, Google-Extended, Bingbot, YouBot, DuckDuckBot. Each is checked against your robots.txt individually.
Do I need an llms.txt?+
Not strictly. But sites that publish a curated llms.txt are easier for engines to summarize and cite accurately. We score whether yours exists and whether the structure is clean.
How does Rook Ranker score the GEO layer?+
441 individual signals across crawl access, llms.txt, freshness, knowledge graph hooks, ingestion friendly structure, citation patterns, and entity reinforcement. Around 163 score on a typical page; the rest light up as you add locale variants and bot directives. Each is weighted by impact.
Be cited. Not skipped.
Score the page generative engines actually read. Free to start.