How to Reverse-Engineer Competitor Pages Cited by ChatGPT

Why does ChatGPT cite competitors whose content better matches the question?

When ChatGPT searches the web, it retrieves pages and then picks passages that answer each part of the question. A competitor wins the citation when their passage fits better, even if your page is stronger overall. Typical reasons:

• The wording matches. Their heading or first sentence uses the same terms as the question, such as "for small teams" or "under $50".
• The answer is explicit. They state the fact directly ("costs $29 per user per month"), while yours implies it or hides it behind "contact us".
• The condition is covered. The question asks about a specific case, such as an integration, a region or a company size, and only their page addresses it.
• The format is easy to extract. A table, a numbered list or a short definition is easier to use than a long narrative paragraph.
• The page is retrievable. Their content is in plain HTML; yours may depend on JavaScript or be blocked for ChatGPT's crawler.

None of these is about domain authority alone. That is why a smaller competitor's page can be cited over a larger brand's.

How do you collect the competitor citations to study?

Start from your own question set rather than from the competitor:

1. Run your buyer questions in ChatGPT with search on, several times each, in clean sessions.
2. Log every cited URL, the question, and the sentence the citation is attached to. ChatGPT places citation markers next to the sentences they support, which tells you what each source was used for.
3. Filter to competitor pages and count how often each is cited and for which questions.
4. Pick the five to ten competitor pages cited most often for questions you care about.

A fuller version of this logging routine is in how to find the pages AI answer engines cite for your questions. For spotting the topics where rivals consistently beat you, see how to find topics where competitors outperform you in ChatGPT.

How do you find the passage ChatGPT actually used?

This is the step most people skip, and it is the most useful:

1. Take the cited sentence from the answer and note its key facts: numbers, names, conditions.
2. Open the competitor page and search it for those facts, not the whole sentence, since the answer usually paraphrases.
3. Mark the passage. Note where it sits on the page (first paragraph, a table, a FAQ, halfway down), what heading it sits under and how long it is.
4. Check whether the answer copied, summarised or changed it. Sometimes the answer adds a detail that is not on the page; that detail came from another source or from memory.

After doing this for ten or so citations, patterns become obvious, for example that most cited passages are the first two sentences under a question-style heading, or rows in comparison tables.

What should you look for on a cited competitor page?

Assess each page on eight attributes and record them in a sheet:

Do the same for your own nearest page. The differences between the two columns are your gap list.

How do you compare your page with the competitor's?

Put them side by side for the specific question they were cited for:

• Does your page answer the question at all? If not, you need a section or a new page.
• Is your answer as easy to find? Move it up and put it under a heading that matches the question.
• Is it as specific? Replace general claims with the facts buyers need.
• Do you cover the condition in the question? Add the case the competitor covers, if it applies to you.
• Can ChatGPT fetch and read your page? Check robots.txt, firewall rules and whether content needs JavaScript.

Also check where the competitor gets mentioned elsewhere. If the page is cited alongside third-party reviews that praise them, the page alone is not the whole story. See how co-citation and co-occurrence affect AI recommendations.

How do you use what you learn without copying?

Copying a competitor's structure and facts rarely wins, because engines have little reason to prefer a near-duplicate. Instead:

• Match their clarity, beat their substance. Answer as directly, then add what they lack: your own data, worked examples, clearer conditions, honest limits.
• Cover the questions around theirs. If they answer "how much does it cost?", you can also answer "what does the price include?" and "what do extra users cost?".
• Use your own evidence. Customer numbers, case results, test data and expert input give engines facts that exist nowhere else.
• Be fair about competitors: a comparison that states where a rival is stronger gives an engine a usable passage for "[Rival] vs [us]" questions, and protects your credibility with the buyer who clicks through.

Then re-run the same questions after the pages are recrawled and compare citation counts with your baseline.

What does a finished reverse-engineering sheet look like?

One row per cited competitor passage:

The rows above are illustrative. Work through the sheet in order of question value, and record the date of each change so you can match it to later results.

What do people ask about reverse-engineering AI citations?

Can I just ask ChatGPT why it cited a competitor?

You can ask, and with search on it can point to the sources it used, but its explanation of why is not reliable evidence. Matching the cited sentence to the page passage is more dependable.

Does this work for Perplexity and Google's AI results too?

Yes. Perplexity shows sources for nearly every answer, which makes passage matching easy. Google's AI Overviews link sources beside the text, and the same comparison applies.

How many competitor pages should I analyse?

Five to ten pages cited for your most valuable questions is usually enough to see the patterns. Add more only if the patterns are unclear.

What if the cited competitor page is weaker than mine?

Then look at retrieval and fit: whether your page is accessible to ChatGPT, whether it answers that exact question early, and whether it uses the buyer's wording.