Last verified: 3 November 2026
ChatGPT recommends a competitor when the pages it finds for the buyer’s question answer that question more directly than yours, and those pages are often not the competitor’s own site but comparison tables, affiliate lists or forum threads that name them. The fix is to find which pages are being quoted and publish a clearer, better-evidenced answer on your own domain.
Key takeaways
- Citation is the last step of a search process, not a judgement about which brand is best.
- Assistants split a question into narrower searches, so a competitor can win on one sub-question you never answered.
- Only 37.9% of pages cited in Google’s AI Overviews rank in the top 10 for the same query (Ahrefs, March 2026), so ranking first does not guarantee a mention.
- Answers vary by mode and over time: only 25.6% of cited domains matched between two ChatGPT reasoning modes for the same prompt (Semrush, June 2026).
Why does the competitor get named instead of you?
Three causes account for most cases we see.
- They answer the comparative question and you do not. Your site says what you offer. A comparison page says which option to choose, which is much closer to what people ask an assistant.
- Third parties speak for them. Ranked lists, review sites and forum threads that mention a competitor give the assistant an independent source to quote. If those pages do not mention you, you are invisible in the answer.
- Your facts are hard to lift. Licence details, prices and process buried in PDFs, behind tabs or deep in long pages are harder for an assistant to quote than a short, direct answer.
It is rarely about how big the competitor is. Our own work for Stelrad showed the opposite case: a manufacturer that already ranked first for “radiators” was mentioned as a product while retailer buying guides were cited as the authority, until the advice content was rebuilt as answers.
How do you find out which pages are being quoted?
Run the buyer questions in your category through ChatGPT, Perplexity, Gemini and Google’s AI features, and record three things for each: whether you are named, which competitors are named, and which URLs are cited. Do it more than once, and in more than one mode, because answers change. A single screenshot is a poor measure.
The cited URLs are the useful part. They tell you exactly which page you need to out-answer, and whether that page is a competitor’s site, an affiliate, a directory or a forum. We cover how to set this up properly in how to track brand mentions across ChatGPT, Gemini and Perplexity.
What should you change?
- Publish the comparison facts you own. Your licence, coverage, pricing structure, process and protections, stated plainly in text and tables.
- Answer the sub-questions. Give each narrower question a direct answer on a page that opens with it.
- Correct the third-party pages. Review the affiliate, comparison and directory pages being quoted about your category and fix outdated facts about you. Gambling operators carry responsibility for affiliates they contract with under licence condition 1.1.2.
- Earn independent corroboration. Digital PR and links pointed at the same answer pages, so your claims are confirmed off-site.
- Do not pay for mentions. Bought AI citations are a shortcut Google is actively acting against, as we set out in Google has declared war on bought AI citations.
What does a good outcome look like?
For Stelrad, it meant 2,389 AI citations across 405 pages and a 62% share of voice in AI Overviews for radiator questions, against 30% for the next retailer, measured in Ahrefs Brand Radar. The competitor did not disappear from the answers. Stelrad’s own pages became the ones the answers relied on.
If ChatGPT is recommending someone else for the questions your buyers ask, talk to our AI visibility team.
