Cause 1: the engines cannot read you
A blanket bot-block, an aggressive WAF rule, a JavaScript-only render with nothing in the HTML - any of these quietly removes you from the retrieval pool. Diagnosis: check robots.txt for AI bot rules and your logs for AI crawler activity. Fix: a deliberate crawler policy and server-rendered content.
Cause 2: you never answered the questions
Buyers ask engines full questions - "which analytics setup do I need for a clinic" - and most of those questions have zero search volume, so keyword-driven content plans never produced a page for them. The engine cannot cite what nobody wrote. Diagnosis: map the real question space (we do this in the audit). Fix: answer-format pages for the questions, volume be damned - in our own tracked sample, that is precisely the content AI engines cite most.
Cause 3: your pages are not quotable
An engine lifts passages, and marketing prose does not lift. A page that opens with "In a rapidly evolving digital landscape..." loses to one that opens with the answer. Fix: stated summary up top, steps, tables, one idea per passage - format as a ranking factor.
Cause 4: nobody corroborates you
Engines weigh consensus. If your brand exists only on your own domain - no directory presence, no third-party mentions, no community answers - the model has one source claiming you matter, and it is you. Fix: earn corroboration where your category's evidence lives: industry lists, communities, comparison content by others.
Cause 5: you are stale in the index
Retrieval favors fresh fetches; a site that has not changed in two years drifts down the pool. Fix: real updates on the pages that answer live questions - and crawler-log monitoring to confirm re-fetching follows.
The order matters: fix access before content, content before format, format before authority. Teams that start with "write more blog posts" usually had cause 1 or 2 all along.