Why ChatGPT Recommends Your Competitor Instead of You

AI gives shoppers a short list, and most brands miss it

Ask ChatGPT for the best running tights for marathon training and you'll get four or five names, a line on each, and a suggestion. There's no page two. If you aren't on that list, the shopper never hears about you.

The list is shorter than most brand owners realise. Our State of AI Visibility study tracked 2,541 ecommerce brands across ChatGPT, Perplexity and Google AI Overviews. In October 2026, 72.6% of them appeared in fewer than 5% of AI shopping answers for their category. The top three brands in a category took 37.4% of every brand mention on average.

So when a founder tells me ChatGPT keeps recommending their competitor, they're usually right. And it's usually the same two or three competitors every time. The useful question is why those brands, and what the assistant is actually reading when it picks them.

Why ChatGPT recommends your competitor: it reads what others say about them

An AI assistant doesn't browse your store the way a shopper does. It answers from what it learned in training plus what it finds when it searches the web at the moment you ask. Both depend on what has been written about your brand, mostly by other people.

What it learned in training matters more than most brands realise. In August 2026, researcher Suganthan Mohanadasan looked at the search queries ChatGPT writes for itself before it searches the web. In 21 of 27 conversations, ChatGPT's first search already included brand names the user never typed. Brands it put in its own search reached the final answer 68.9% of the time. Brands it only came across in the pages it read made it 2.1% of the time. It was a small study, 57 conversations from one account and mostly software categories, but the pattern is clear. ChatGPT often has a short list in mind before it looks anything up, and that list comes from what the web has said about brands over time.

OpenAI's own documentation points the same way. It says products are tailored to each shopper based on preferences, context and product data, and that reviews, rich media and performance signals in a product feed improve ranking and relevance.

So where does the outside information come from? The citation studies keep pointing to the same kinds of sources.

• Forums and communities. Peec AI analysed 30 million sources cited by ChatGPT, Gemini, Perplexity, Google AI Mode and AI Overviews in the US. Reddit was the most-cited domain overall, and the first or second most-cited on every platform. Real people comparing real products is exactly what an assistant wants to quote.
• Review sites. The same analysis found review platforms like Yelp in the top 10, with Trustpilot and Tripadvisor in the top 100.
• Editorial roundups. A "best running tights of 2026" article from an established publisher carries weight because it's independent.
• Search rankings. When ChatGPT searches live, it reads pages that rank. If you're invisible in search, you're harder for it to find.

Your own claims about yourself aren't on that list. Your product pages matter, but mostly as a source of facts. "The world's most comfortable tights" on your own site carries little weight. A runner on Reddit saying your tights didn't chafe over 30 kilometres carries a lot.

Big brands don't win by default

The natural assumption is that AI just recommends the biggest names. The data says otherwise, and ChatGPT is where it's most obvious.

Ahrefs looked at how closely the number of web pages mentioning a brand tracks its visibility in AI answers, using mid-2025 data. For Google AI Overviews the link was strong, with a correlation of 0.65. For Perplexity it was weak at 0.30. For ChatGPT it was very weak at 0.15. Google leans toward big, established brands. ChatGPT leans on them much less.

Our own Found data backs this up. Glossier appears in just 2.7% of US beauty answers and ranks 41st of 66 brands. Plenty of brands a fraction of Glossier's size sit above it.

That's good news for mid-size brands. On ChatGPT, being clearly described as the best option for a specific need can beat being the brand everyone has heard of.

You earn coverage one market at a time

The same brand can be the default answer in one country and missing in the next. Antler appears in 45.1% of UK luggage answers and 35.4% in Australia, but just 1.9% in the US. Bellroy ranks third in Australian handbags and accessories and 17th in US luggage.

The likely reason is coverage. AI answers draw on reviews, retailers and press in each market. A brand with deep coverage at home and thin coverage overseas gets named at home and skipped overseas. If you sell internationally, check each market separately. Your home result tells you nothing about the others.

Reviews do more work than you'd think

Reviews are one of the few signals OpenAI names directly. They also do something product copy can't. They describe your product in your customers' words, which tend to be the same words shoppers use when they ask an assistant.

A few things make reviews more useful to an AI assistant.

• Volume across more than one place. Reviews on your own store help. Reviews on independent platforms, retailer listings and Google help more, because they're harder to fake. ChatGPT shows aggregated star ratings that may not match any single site.
• Specifics. "Love it!" tells an assistant nothing. "Fits true to size, kept me warm at 5am in July" tells it who the product is for.
• Recency. A product whose last review was in 2023 looks like a product nobody buys anymore.

If you've treated reviews as a conversion tool for your product pages, they're now a visibility tool for every AI answer in your category too.

Your product data is the one part you control

Everything above is about what other people say. Product data is what you say, and you can fix it this month.

OpenAI says Shopify and Etsy catalogues are already integrated into ChatGPT shopping, with no application needed. ChatGPT's merchant program is US-only for now. Shopify's agentic storefronts also put your products in front of buyers inside ChatGPT, Google AI Mode and Gemini, and Microsoft Copilot. So these assistants already read your titles, descriptions, prices, variants and stock levels. The question is whether those say anything useful.

OpenAI tells merchants that richer data, like images, pricing and reviews, helps products surface with more accurate information when shoppers compare options. In practice, that means plain facts over brand language.

• What it's made of, what it's for and who it suits
• Sizing, fit and dimensions written out in words, not only in a chart image
• Price and availability that match what's actually on your site
• The same product names across your store, retailers and marketplaces

An assistant answering "waterproof hiking boots for wide feet under $300" can only recommend you if your product data says waterproof, wide fit and the price. A description that's all mood and no facts gives it nothing to match.

Be the obvious answer to "best for what?"

AI assistants rarely answer "what's the best brand" in the abstract. They answer "best for marathon training", "best gift under $100", "best for a small apartment". They name the brands the web keeps connecting to that specific need.

The leaders in Found's lists tend to have one clear position that gets repeated everywhere. Reviewers, press and customers all describe them the same way. A brand the web describes five different ways doesn't become the answer to any of them.

Here's how the signals compare.

Only one of those is fully in your hands. That's why the brands winning AI search treat it as a reputation job as much as a website job.

What to be sceptical of

AI search has attracted a lot of quick-fix promises. A few deserve a hard look.

• Paying your way in. ChatGPT started showing ads in the US in February 2026, for logged-in users on the free and Go tiers. OpenAI says ads don't shape the answers themselves. An ad might buy you a spot next to the answer. It won't put you on the list inside it.
• Writing your own "best of" lists. Publishing "the 10 best brands in our category" with yourself at number one is tempting. AI tools lean on independent sources for a reason, and a list where the author ranks itself first is the opposite of independent.
• One screenshot. In 35 of the 56 categories Found tracks, ChatGPT, Perplexity and Google AI Overviews name a different number one brand. Top-10 brands swing an average of 4.2 points between weekly scans. One check on one platform tells you very little.
• Tools that promise a ranking. Nobody controls what ChatGPT says. Anyone guaranteeing a specific position is guessing.

The real cost is time and ownership. Coverage, reviews and a clear position build over months, and someone in your business has to own them. Expect a few weeks just to get a reliable baseline, and longer before the answers shift.

Where to start

Start by finding out where you actually stand.

1. Check your category on Found, our AI visibility ranking. It runs the same 150 shopping questions for each category every week, so you can see who gets named and how often.
2. Ask the questions your buyers ask. Pick ten real questions for your category and run them on ChatGPT, Perplexity and Google. Note who gets named, and which sources the answers point to. Those sources are your coverage targets.
3. Fix your product data. It's the fastest win and the one you control. Read your top 20 product descriptions as if you were an assistant trying to answer a specific question.
4. Go after the sources that keep appearing. If the same three review sites and two publications keep showing up in answers for your category, that's where you need to be.

Our guide to getting your content cited by ChatGPT and Perplexity covers the content side. Once you're on the list, measuring whether AI search is making you money is the next step.

If you want help with the product data and content side, see our Shopify AEO service. If you'd rather have it done for you end to end, Lift, our 100-day AEO partnership, starts with exactly this baseline and works through the gaps.

The bottom line

Your competitor isn't getting recommended by accident. The places AI assistants read describe them well and often, and their product data backs it up. You can close that gap. It takes longer than a website update, and it starts with knowing where you stand today.