What you will learn
- Why a site can rank for Google and still be invisible to assistants.
- What has to live in the first HTML response for a passage to be quotable.
- How third-party lists sit alongside owned pages in a visibility programme.
What was actually wrong?
Nothing on the site was broken for a human being, which is exactly what made it hard to diagnose.
Practice area detail loaded when a visitor clicked a tab, the case studies were PDF downloads, and the team directory was assembled in the browser from a JSON file. A person clicking around would have said the site was perfectly good, and the firm's own agency had been reporting healthy rankings for two years.
The tabs and the directory were empty to any crawler that does not execute JavaScript. The PDFs were readable, and still not what those crawlers cited for the firm's core claims.
Why did assistants ignore it?
Because there was nothing they would quote. GPTBot, ClaudeBot, PerplexityBot and Bingbot do not execute JavaScript, so a tab that fetches its own content is an empty container to them, and a PDF is not somewhere they look for a firm's core claims. The practice ranked well in Google, which renders, and appeared nowhere in the assistants their clients were increasingly asking first.