The Answerability framework.
The definitions the rest of the practice is built on — what we measure, and why a citation depends on all three.
Answerability is the composite a company earns when an AI system decides whether to name it: the ability to be retrieved, trusted, and used as an answer to a buyer's question. It is earned across three pillars — Content, Retrieval, and Trust — and it is constrained by the weakest of them, not their average. A page can clear two pillars and still never be cited because it fails the third. We do not abbreviate the framework to an acronym; we spell out the pillars.
How a citation is earned
Before an engine names you in an answer, your page has to clear three gates in order. Read it as four questions: Do you have the content? Can the engine retrieve it? Does it trust the source? — and only then, are you answerable? Miss any one gate and the engine moves on to a competitor.
Do you have the answer?
Content that addresses what buyers actually ask, in a form an engine can lift.
Can AI reach it?
Your retrieval surface — the slice of your site engines can crawl, parse, and structurally understand.
Will AI vouch for it?
When several sources could answer, the engines treat yours as cite-worthy — evidence plus outside corroboration.
A page can clear two gates and still fail the third — so your weakest pillar, not your average, decides the outcome. The hard part is rarely the fix; it's knowing which gate is closed.
The three pillars, defined
Content
Is there content that answers what buyers actually ask, in a form an engine can lift — coverage plus answer-shape?
Retrieval
Can the engines access, crawl, parse, and structurally understand that content in the first place?
Trust
When several sources could answer, do the engines treat yours as cite-worthy — internal evidence plus external corroboration?
Two companies, the same problem, different gates
These are illustrative — but they are the two patterns we see most. Both companies are nearly invisible to AI; the reason is opposite, and so is the fix. The composite alone won't tell you which. The per-pillar read does.
Bottleneck: Trust. The pages are crawlable and answer the question well — but the engines don't yet treat the company as a source they'll stand behind, so a better-corroborated competitor gets cited instead. The work is evidence and outside corroboration, not more content.
Bottleneck: Retrieval. A well-known name with real authority — but the answers live behind JavaScript the crawlers don't render, or in PDFs they don't parse. The engines never see the content to cite it. The work is the retrieval surface, not reputation.
The canon
The working primer defines the category; each pillar note is the long-form, dated reference for one axis, with real cross-engine captures. These are written to be quoted — by readers and by machines.
Generative Engine Optimization: a working primer
A working definition of GEO and the Answerability framework (Content, Retrieval, Trust) — a citable reference, not a marketing summary.
Content: the first pillar of Answerability
Whether you have content that answers what buyers actually ask, in a form an engine can lift — with real cross-engine captures.
Retrieval: the second pillar of Answerability
Whether AI systems can access, crawl, and parse your content — JavaScript failures, crawler policy, structured data — with a real AI-crawler audit.
Trust: the third pillar of Answerability
When an engine has several sources, why it cites one over another — internal evidence plus external corroboration, with a real entity-graph audit.
How we apply it
The framework is the lens; two companion pages show it at work. The methodology is how the pillars are measured against a standing prompt set across five engines; the glossary fixes the vocabulary the practice uses consistently.
From definitions to evidence. The framework says what to look for; the Answerability Index shows it in real cross-engine data, sector by sector. To see it run on your own company — which gate is closed, and the work order to open it — that's how we work.
Find out which gate is closed for you.
The Diagnostic scores every cited URL on Content, Retrieval, and Trust across all five engines — then hands you the work orders, an Excel playbook, and two deploy-ready pages to start opening it.