8 Reasons AI Assistants Aren't Recommending Your Website
When an AI assistant never mentions a website, it is almost never one mysterious problem — it is usually one or more of a small, consistent set of fixable issues. These are the same eight signals RankLLM checks on every scan, because they are what actually determine whether ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews treat a site as worth recommending.
1. AI crawlers are blocked
The single most common and most decisive issue. If GPTBot, ClaudeBot, Google-Extended, or PerplexityBot are disallowed in robots.txt, or a page carries a noindex tag, that platform cannot recommend you — full stop, regardless of content quality.
2. Content never directly answers a question
AI systems favor content shaped like a clear question followed by a direct, short answer. Pages that only cover a topic in flowing narrative prose, without ever explicitly stating the answer to an implied question, give an AI nothing clean to lift.
3. No visible sign of who is behind the site
An anonymous site is a risky recommendation. Without a real About page, contact information, or visible authorship, AI systems — especially more cautious ones like Claude — tend to prefer a source they can vouch for.
4. Writing is too dense to summarize confidently
Long sentences, heavy jargon, and buried key points make it harder for a model to summarize a page accurately, which makes it a riskier source to quote. Clear, plain writing gets lifted more often simply because it is safer to lift.
5. The topic is covered too thinly
A couple of shallow paragraphs rarely earns a citation when a competing page thoroughly answers the question and its natural follow-ups. Depth is a real, measurable factor — not padding, genuine coverage.
6. No structured data (schema/JSON-LD)
Without machine-readable labels, an AI has to infer what a page is about from formatting and text alone. Structured data removes that guesswork and is disproportionately valued by Google AI Overviews specifically.
7. Content looks outdated or undated
No visible publish or update date leaves an AI unable to judge whether information is still accurate — and it will usually choose a visibly fresher competitor over an ambiguous one, especially for anything time-sensitive.
8. Technical friction slows or blocks reading
A slow server response, a missing title tag, a broken canonical URL, or a page that requires heavy JavaScript to render any content at all can each make a page harder for an AI crawler to fetch and parse cleanly than a competitor's page.
Frequently asked questions
Which of these eight matters most?
Crawler access is the hard gate — if it fails, nothing else matters for that platform. After that, it varies by platform: Claude weighs trust most heavily, Perplexity weighs freshness and direct answers, and Google AI Overviews weighs structured data most heavily.
How do I check which of these apply to my own site?
RankLLM runs a free scan that checks all eight signals against your actual page in under a minute and explains each one in plain English, with the technical detail available if you want it.