A competitor's page owns the AI citation: what to check before you write

You run your key prompts and the same URL keeps appearing in the sources. It's a competitor's guide, or a comparison page, or a glossary entry. It is cited again and again, and your equivalent page is nowhere. The pressure to "write something better" arrives within the hour.

Resist it for a day. Most of the time the page that owns a citation owns it for a reason you can identify, and the reason isn't always quality. Writing before you understand it gets you a longer page that loses in the same way.

Step 1: confirm it's a pattern

One citation in one answer is noise. Check that the page appears across several prompts and several runs. If it's cited in most answers for a cluster of related prompts, it owns that cluster. If it shows once, you're looking at a coincidence.

Write down which prompts it covers and which engines cite it. A page cited by one engine only often depends on that engine's own index or partner sources.

Step 2: read the page as the engine would

Open the cited page and ignore the design. Look at structure and what is stated plainly:

• Does it answer the question in the first few lines, or bury it?
• Does it use headings that match the sub-questions people ask?
• Does it include specifics: numbers, steps, a table, named examples?
• When was it last updated, and does it say so?
• Is it on a site with a track record in the topic, or is it a one-off?

Then compare your equivalent. Typical gaps look like this: the competitor states a definite answer, you hedge; they cover a sub-question you skip; they've updated, you haven't. Those are observations about individual pages, not guarantees about what an engine rewards.

Step 3: check what the engine actually asked

Many AI answers are built from several searches the engine runs behind the scenes, often called query fan-outs. A competitor's page may win not because it's the best answer to your prompt, but because it answers one of the fan-out queries the engine generated. If you only target the prompt, you miss the sub-queries.

Our post on query fan-out explains the mechanism. Look at which sub-queries your competitor's page matches and whether any page of yours does.

Step 4: decide between refresh and new page

Not every gap needs a new article.

• If you have a page on the topic that is out of date or vague, refresh it. See our post on rewriting existing pages for citations.
• If you have nothing that addresses the sub-question, write a new page.
• If the competitor wins on a source type you can't match (a government resource, a large publication), the better move may be to get covered there than to compete head-on.

The content gap guide lays out how to list the missing answers systematically. Be honest in the list. "We have nothing on this" is more useful than "we could improve this".

Step 5: write from a brief, not from a mood

When you write, start with a short brief: the prompt cluster, the sub-questions, the competitor pages that are cited, what each one states, what you can say that is true and they don't. Original material is your advantage: your own data, your own process, your own examples. Don't rewrite their page with your logo on it.

Then measure. Check back after a few weeks to see if the page gets crawled and cited. We wouldn't promise a result. Citations can take time to move, and some won't.

Where Promptwatch fits

The manual workflow has a lot of copying between tabs: prompts, sources, fan-outs, notes, then a brief in a doc. The platform we'd use for this is Promptwatch, because it keeps the diagnosis and the draft in the same place.

On the diagnosis side it shows page-level citations and query fan-outs per model, and content gap analysis with coverage scores. The Page Tracker lets you follow specific URLs, yours or a competitor's, so you can see whether that competitor page is cited steadily or sporadically.

On the writing side, the Promptwatch changelog describes how the Content Agent changed on October 12. It no longer needs a tracked prompt. You can choose Describe a topic, or Target a tracked prompt to use its query fan-outs and the competitor pages AI engines already cite. You can also click Create content on an Answer Gap task, a query fan-out or a cited page, which opens the content agent with that context filled in. Add your own competitor URLs as sources, up to five per article. Content is written in the language of the target prompt or monitor, and finished drafts include a card showing how the content was created: the inputs, research and internal links behind the article.

That chain fits the steps above: you find the cited page, click Create content from it, and the brief is mostly assembled. The Content Agent also looks for relevant pages on your site before writing, so internal links go into the first draft.

Plans, plainly

The scheduled, automated Content Agent that plans, writes and publishes against gaps on a calendar is a Business and Enterprise feature. Business costs $579 a month. If you're on a lower plan, the platform still has content gap analysis and on-demand content creation, but the plan limits differ, so check the pricing page before relying on it. We won't guess at those limits here. Essential and Professional customers shouldn't assume the automated agent is part of what they bought.

What this doesn't do

It doesn't guarantee the new page will be cited. It doesn't replace editing. The agent can produce a draft grounded in the gap and the sources, and someone who knows the product still has to check claims, add examples, and decide whether the page deserves to exist.

There is also a quieter risk. If five competitor URLs are your only source material, your draft will resemble theirs. Feed it your own material as well.

The short version

Confirm the pattern, read the page like an engine, check the fan-outs, choose refresh or new, and write from a brief. If you can't say what's missing from your site in one sentence, you aren't ready to write yet.