Why AI Search Recommends Your Competitor Instead of You?

The message usually arrives with a screenshot. Someone asked ChatGPT for “the best [your category] for small teams,” and the answer listed three companies. One of them is the competitor you beat on price, on features, on support tickets, on everything you can measure. You weren’t in the list at all.

It stings. And the first reaction is almost always the wrong one: publish more blog posts, add some schema, maybe an llms.txt file, and hope. Before you fix anything you need to know which problem you actually have, because there are about seven different reasons AI search picks a competitor over you, and they need different fixes. This is the diagnosis part. We wrote the repair side separately.

Key takeaways
  • One bad answer proves nothing. AI tools return the same brand list less than 1 time in 100, so test the same prompt 20 or more times and count how often you appear.
  • AI picks brands from two places: what the model learned in training, and what its live search finds for several rewritten versions of the question.
  • Off-site mentions matter more than anything on your own site. Reddit threads, YouTube and web mentions all correlate with AI visibility at around 0.64 to 0.75.
  • The most common reasons are simple: you’re blocked from the crawler, you’re missing from the “best of” lists, or your site never says plainly what you do and who it’s for.
  • Skip the quick tricks. Google says it doesn’t use llms.txt or special schema for its AI features, and page count barely correlates with AI mentions.

First, check that it’s a pattern and not one unlucky answer

This part gets skipped constantly, so let’s start here. AI assistants don’t keep a fixed ranking the way Google keeps ten blue links. They generate a new answer every time.

SparkToro and Gumshoe had 600 volunteers run 12 recommendation prompts through ChatGPT, Claude and Google’s AI almost 3,000 times. Their finding: there’s less than a 1 in 100 chance that ChatGPT or Google’s AI gives the same list of brands twice, and roughly 1 in 1,000 that the order matches too. Some answers had 2 or 3 names. Others had 10.

So your screenshot might just be a bad roll. Or it might be a real gap. The way to tell is boring but it works:

  1. Write 5 to 10 prompts a real buyer would type. Not your brand name. Things like “best invoicing software for freelancers in Germany” or “alternatives to [competitor].”
  2. Run each one at least 20 times across ChatGPT, Perplexity and Google AI Mode. Use fresh chats, logged out where you can.
  3. Count how often you appear, and how often each competitor appears. Ignore position. Position is noise.

If you show up in 40% of answers and your competitor shows up in 70%, you have a gap to close. If you show up in 0 out of 60, that’s a different and more urgent problem, and it usually points to reason 1 or reason 5 below.

How AI search decides which names to mention

Two things feed every answer. The first is memory: what the model absorbed about your category during training. If the web talked about your competitor a lot before the training cutoff, the model already “knows” them. You can’t edit that directly.

The second is live search. When a question needs current or specific info (and “which tool should I buy” usually does), the assistant goes and searches. Google calls this query fan-out and describes it as a set of concurrent, related queries the model generates to fetch more results. So “best CRM for a dental clinic” quietly becomes five or six searches: CRM for dentists, dental practice software reviews, CRM pricing comparison, and so on.

Your competitor gets named when they show up in enough of those results, and in enough of the pages those results point to. That’s the whole game, really. The seven reasons below are just the different ways that can fail.

The 7 reasons AI recommends your competitor

1. The crawler can’t read you

Embarrassing, but common. We’ve seen it more than once on sites where a security plugin or a CDN rule blocks unknown bots. OpenAI is blunt about it: sites that opt out of OAI-SearchBot will not be shown in ChatGPT search answers, and after you fix robots.txt it takes about 24 hours to update. Google has its own rule. A page has to be indexed and eligible to show with a snippet, so a stray nosnippet tag can take you out of AI Overviews and AI Mode completely.

Check robots.txt, your firewall logs, and your Bing Webmaster Tools coverage. If you’re at 0 out of 60 answers, start here.

2. The web talks about them, not you

This is the big one. Ahrefs looked at 75,000 brands across ChatGPT, AI Mode and AI Overviews and found branded web mentions correlate at 0.66 to 0.71 with AI visibility. YouTube mentions came out even higher, around 0.74. Those are correlations, not proof, but the direction is hard to argue with.

Your competitor might have a weaker product and a louder footprint. Podcast appearances, a founder who posts everywhere, a couple of press mentions, a YouTube reviewer who liked them. The model sees their name next to your category hundreds of times, and yours maybe a dozen.

3. They’re in the “best of” lists and you aren’t

When the assistant runs “best X for Y,” what comes back is mostly listicles and comparison pages written by third parties. If your competitor is in 9 of the top 10 “best” articles for your category and you’re in 1, the answer basically writes itself. Google a few of your buyer prompts yourself and note who appears in the lists. It’s quick and it explains a lot.

4. Reddit and YouTube never heard of you

Ahrefs ran a second study on mention types. Reddit mentions, in threads that rank on Google, had the strongest correlation of all, about 0.75, ahead of news articles and TikTok. The typical brand in their sample had only 8 Reddit threads mentioning it, and 29% had none.

Peec’s analysis of 30 million AI citations puts Reddit first, then YouTube, LinkedIn and Wikipedia, with G2 and Yelp also in the top ten. If you don’t exist in the places AI reads most, you’re asking it to recommend a stranger.

5. Your site never says what you are in plain words

A lot of homepages say things like “empowering teams to unlock growth.” Nice. But what is it. A CRM. For whom. Dental clinics with 2 to 20 chairs. In which country. How much does it cost.

The AI is matching your pages to a fan-out search like “CRM for dental clinics pricing.” If none of your pages contains those words in a clear sentence, you can’t be retrieved for it, no matter how good the product is. Your competitor wins simply because their page says “CRM for dental practices, from $49 a month.”

6. Your facts don’t match across the web

Old pricing on a review site, a different company name on a directory, a feature list from 2024 on G2. When sources disagree, the model either picks the wrong version or plays it safe and recommends someone whose story is consistent. We covered this in more detail in what to do when AI describes your business incorrectly.

7. Their pages answer the buying question and yours don’t

Buyers ask AI comparison questions. “X vs Y.” “Is X worth it.” “Cheaper alternative to X.” Your competitor has a pricing page with real numbers, a comparison page, maybe an honest “who it’s not for” section. You have a “contact sales” button. The model can only quote what exists, so it quotes them.

Quick diagnosis table

What you seeMost likely reasonFirst thing to check
You never appear, not even for your own niche1 or 5robots.txt, firewall, Bing indexing, homepage wording
You appear sometimes, competitor appears most of the time2, 3 or 4Who’s in the top “best X” lists, Reddit and YouTube search for both names
You appear in ChatGPT but not in AI Mode (or the reverse)1 or 3Google snippet settings, Bing vs Google rankings for fan-out queries
You’re mentioned, but with wrong prices or features6Review sites, directories, old press
You show up for “what is X” but not “X vs Y”7Do you have pricing and comparison pages at all

What isn’t the reason (so don’t spend a week on it)

A few things get blamed a lot and probably aren’t your problem.

  • Missing llms.txt. Google says it doesn’t use it, and that these files will neither harm nor help your visibility in Google Search. It’s fine to have one. It won’t move this.
  • Missing special schema. Same guide: there’s no special markup for AI search. Structured data still helps with rich results, it just isn’t the switch.
  • Not enough content. In the Ahrefs data, number of site pages correlated with AI visibility at only about 0.19. Fifty more thin posts won’t beat one competitor who’s in every buyer’s guide.
  • The AI being “paid off.” Ads now appear inside some AI answers, but the organic recommendation list isn’t for sale, and the data points to mentions, not ad spend.

Where to go from here

Run the 20-times test first. Honestly, it changes the conversation, because “ChatGPT ignores us” becomes “we show up in 12% of answers and they show up in 64%,” and that’s something you can track and improve.

Then match your result to the table. Reasons 1, 5 and 7 are on your own site and you can fix them this week. Reasons 2, 3, 4 and 6 are off-site and take a few months, because you’re changing what the rest of the web says about you. That’s slower, but it’s also where the biggest gap usually is. When you’re ready for the actual repair work, our 4 fixes for when AI recommends your competitor walks through it step by step.

FAQ

Why does ChatGPT recommend my competitor and not me?

Usually because the web mentions them more often next to your category, they appear in more “best of” lists, or your own pages don’t clearly say what you do and for whom. Sometimes it’s technical, like a blocked OAI-SearchBot.

How do I check if AI search recommends my brand?

Run 5 to 10 real buyer prompts at least 20 times each in ChatGPT, Perplexity and Google AI Mode, then count how often you appear versus your competitors. A single answer means very little because the lists change almost every time.

Does Google AI Mode use the same signals as ChatGPT?

Not exactly. They search different indexes, so you can show up in one and not the other. But brand mentions correlate strongly across all of them, and Ahrefs found AI Mode and ChatGPT mention largely the same brands.

Will adding llms.txt or schema make AI recommend me?

Unlikely. Google says its AI features don’t use llms.txt and don’t need special schema. Crawl access, clear pages and third-party mentions matter far more.

How long does it take to start appearing in AI answers?

On-site fixes like unblocking crawlers or rewriting a vague homepage can show up within days to a few weeks. Earning mentions in lists, Reddit and reviews usually takes two to four months to show in the numbers.

Disclosure: information and opinion, not sponsored. No affiliate links. Studies and documentation were checked on November 10, 2026. AI answers change week to week, so run your own tests before you act.

Not sure which reason is yours?
Our free AI visibility checker shows whether ChatGPT and Perplexity mention your brand, and who they recommend instead.
Check your AI visibility