ChatGPT recommends a competitor instead of us: why and what to do

ChatGPT recommends your competitor instead of you because, for that buyer question, more of the sources it trusts name your competitor: "best X" lists, review sites, comparison articles and community threads. To close the gap, find the sources cited for that question, map where the competitor appears and you don't, then earn accurate coverage there and publish pages that answer the question directly.

If you are missing from every answer, not just losing to one rival, start with why ChatGPT doesn't mention your brand.

01Is the competitor really winning, or is it noise?

AI recommendations vary from run to run. SparkToro and Gumshoe found that the same prompt rarely returned the same list twice, but some brands still appeared in 60% to 90% of responses for a given intent.^[1] That second number is what you are competing on.

So measure it properly: run each buyer question many times, across engines and markets, and count how often each brand appears. Use the wording buyers actually type, taken from sales calls and search data, not your internal category name. Our guide to AI share of voice explains how to turn that into a metric you can track. If your competitor appears in most runs and you appear in a few, the gap is real.

02Why does AI favor one brand over another?

Answer engines assemble recommendations from what they find. When they search, they lean on pages that already compare options, and those pages have a strong pull.

Ahrefs studied 26,283 source URLs behind ChatGPT answers to 750 recommendation prompts and found "best X" blog lists were 43.8% of all cited page types. Brands placed higher in those lists were more strongly associated with being recommended.^[2]

Third-party presence beyond lists matters too. In SE Ranking's study of 129,000 domains, those with profiles on several review platforms averaged 4.6 to 6.3 ChatGPT citations versus 1.8 without, and domains with very heavy Reddit discussion averaged 7 citations versus 1.8 for minimal presence.^[3]

03How do you find which sources tip the answer?

Work from the prompt outward. For each buyer question where you lose, record the sources the engine cites, then search for the same question yourself and list the pages that rank or appear in AI summaries.

1. 1
   Pick the losing prompts

   Choose the 10–20 buyer questions where the competitor appears far more often than you.
2. 2
   Collect cited sources

   Run each prompt several times on each engine; save every cited URL and note repeats.
3. 3
   Map presence

   For each source, mark: competitor named? You named? Position? Facts accurate?
4. 4
   Classify the gap

   Missing from a list, outdated info, weak review profile, no comparison page, no community presence.
5. 5
   Prioritize

   Start with sources cited most often across prompts and engines.

Many teams find that a handful of pages carry a large share of the answers. When that happens, the problem is smaller than it first looks: a focused effort on those pages can matter more than a broad content program.

Note the accuracy of each source as you go. A list that names you with a discontinued product or the wrong price can hurt more than a list that leaves you out.

04How do you close the gap legitimately?

There are four levers, and they work best together.

Get accurately included where buyers compare. Contact publishers of lists that cite your category and offer current facts, pricing and a contact for review. Some lists are self-published by vendors; Peec AI found about 11% of citations in its 232,000-citation study came from self-promotional listicles.^[4] You can publish your own comparison too, but rank competitors fairly and disclose that it is yours.

Strengthen review profiles. Ask real customers for detailed reviews on the platforms that matter in your category. G2's own analysis of 30,000 AI citations found the link between review volume and citations was real but small,^[5] so treat reviews as one input, not the answer.

Publish the pages the question needs. If buyers ask "best X for regulated industries," you need a page that answers exactly that, with evidence. Comparison pages ("us vs competitor") are legitimate if they are accurate and current.

Show up where practitioners talk. Answer questions in communities with a disclosed company account and real expertise. Our guide to the sources AI trusts covers which ones carry weight.

The backfire column carries legal risk, not just reputational risk. The U.S. FTC's 2024 rule prohibits fake reviews and misrepresenting company-controlled review sites as independent.^[6]

05Does the gap look different on each engine?

Often, yes. Each engine has its own retrieval and source preferences, so a competitor can dominate ChatGPT and be weaker in Google's AI features or Perplexity.

Peec AI's listicle study illustrates the difference: self-promotional lists made up about 3.6% of citations on ChatGPT but about 10.3% on Google AI Mode and 10.4% on Perplexity.^[4] A competitor that relies heavily on its own lists may therefore look stronger on some engines than others.

Google says its AI features use a "query fan-out" technique, running multiple related searches across subtopics, and that pages need to be indexed and eligible for a snippet to appear as supporting links.^[7] That makes classic search visibility for related subtopics part of the picture there.

Measure each engine separately, and fix the sources each one cites. A single blended score hides where the competitor is actually winning.

06What if the AI says something wrong about you while recommending the rival?

Sometimes the competitor wins partly because the answer misstates your facts: an old price, a missing feature, a market you no longer serve. Treat those as accuracy errors first, because no amount of new coverage helps if the engine keeps repeating a wrong reason to exclude you.

Log each error with its likely source and correct it there. The workflow is in how to fix wrong AI information.

07Should you attack the competitor's position?

No. Disparaging content tends to be ignored by engines or, worse, retrieved and summarized in ways that associate your brand with negativity. Compete on clarity and evidence: say plainly who you are best for, back it with proof, and make sure the sources engines read say the same.

08How long does it take to change the recommendation?

It depends on whether answers are search-based or drawn from model memory, and on how quickly third parties update. Corrected sources need to be recrawled, and memory-based answers wait for new model versions. Set expectations in terms of work shipped and repeated measurement, not a fixed date.

09Who should own competitor gaps?

Competitor gaps cross teams: product marketing for positioning, PR for lists and press, customer marketing for reviews, SEO or web for pages and access. Without one owner, each team optimizes its piece and nobody watches the answer.

If you want that loop run continuously, OwnlyReal measures share of answer against named competitors and builds off-site authority through your approvals; see why teams use it.