Skip to main content
If your domain gets no citations in AI answers, the cause is almost always one of the following. They are ordered from most to least common — work down the list.

​
1. You have nothing retrievable to cite

The most common cause by far. Answer engines cite pages that answer a specific question factually. Most sites have only marketing pages and a blog, neither of which is shaped to be extracted from. A landing page argues that you are good. It does not state what something is, what it costs, how it works, or what its limits are. There is nothing for a model to lift. Fix: publish factual, narrow pages — one topic each, definition in the first sentence. Documentation and help-centre content is consistently the most-cited vendor page type.

​
2. Your pages are not crawlable or not renderable

If body text requires JavaScript execution, many retrieval crawlers will not see it. Content behind auth, behind a cookie wall, or blocked in robots.txt is invisible. Fix: confirm body content is present in the served HTML. Check robots.txt does not block the pages you want cited. Do not gate documentation behind a login.

​
3. Your pages are too broad

A page covering ten topics competes for none of them. Models retrieve to answer a specific question, and a narrow page that answers it precisely beats a comprehensive guide that mentions it. Fix: split. One page, one thing. what-is-x outperforms the-complete-guide-to-x.

​
4. Your content is persuasive rather than factual

Marketing language gives a model nothing to extract. Sentences that could sit on any competitor’s site carry no information. Fix: state facts, numbers, limits, and constraints. Say what the product does not do. Cited pages read like specifications, not brochures.

​
5. The prompts you track are structurally unwinnable

Some prompt clusters are dominated by one platform’s own properties, or by large review sites. In those, third-party content has a hard ceiling regardless of quality. Fix: measure those clusters separately and stop spending content effort against them. See prompt monitoring.

​
6. Not enough time has passed

Pages must be crawled, indexed, and then retrieved before they can be cited. Weeks to months, not days. Fix: wait, and judge by trend across a quarter.

​
7. Your brand is not in the model’s priors

If competitors are named without any page being cited, they are in the training data and you are not. No amount of your own content changes that directly. Fix: accept the timeline. Third-party presence over many months is the only path. See competitor share-of-voice.