SEO
Whether conventional search engines can crawl, parse, and rank the catalogue.
Engage →Generative engine optimisation
GEO
Whether generative engines will cite this store when composing a recommendation.
What it costs to fail
A disallowed agent is a hard exclusion, not a ranking penalty — the store is absent from the answer rather than low in it, and no amount of catalogue quality compensates. Where there is no vendor and no date, a model has nothing to attribute a claim to and prefers a source that does.
What is scored
llms.txt presence and quality, per-agent policy for GPTBot / ClaudeBot / PerplexityBot / Google-Extended, quotable claim density, named entities, dates, vendor and provenance attribution, freshness (updated_at).
Each of these is a published criterion with a published weight — the full rubric.
What fixes it
Publishing llms.txt states your terms to generative crawlers instead of leaving them to infer. Naming a vendor and keeping updated_at current give a model something to attribute the claim to and a reason to prefer the record over a stale one. Unblocking an agent in robots.txt is the one change that removes a hard exclusion rather than improving a ranking.
The other three
Every scan runs all four, in every tier. A catalogue that clears this pillar can still be invisible to the systems the others cover.
