Why ChatGPT Recommends Some Brands and Not Others
ChatGPT recommends some brands and not others based on two things: what its crawler can read and what other sites say. It runs a web search, pulls the pages it can reach, and names the brands that keep showing up across them. No brand can pay for a spot. For best results, earn mentions on the sites it already cites.
Key Points
- ChatGPT finds brands through search, so a store that blocks its crawler does not get recommended.
- Mentions on other sites predict a recommendation better than authority does: 0.664 against 0.266 in Ahrefs’ study of 75,000 brands.
- Wikipedia, Reddit, and large publishers supply most of what ChatGPT cites, and that mix changes without warning.
- Product recommendations run on feed data such as price, availability, and reviews.
- OpenAI says product results are not ads, so the spot has to be earned.
How Does ChatGPT Pick the Brands It Names?
ChatGPT picks the brands it names by searching the web, not by remembering them. OpenAI’s crawler documentation says OAI-SearchBot is “used to surface websites in search results in ChatGPT’s search features,” and that sites opted out of it “will not be shown in ChatGPT search answers.”
That search leans on Bing more than on Google. Seer Interactive compared the two and found 87% of ChatGPT search citations matched Bing’s top organic results, against 56% for Google. A store that has never checked its Bing rankings may be missing from the index ChatGPT uses most.
Bing Webmaster Tools is the quickest way to find out. Add your store, submit the sitemap Shopify already generates, and look at which pages Bing has indexed. Most store owners I talk to have never opened it. Google’s index counts too, even though you don’t need to rank in Google to show up in ChatGPT.
From there, an answer gets built in four stages, and a brand can drop out at any of them:
| Stage | What ChatGPT does | Where a brand drops out |
|---|---|---|
| 1. Rewrite | Turns the shopper’s question into several searches | Your pages do not use the shopper’s words |
| 2. Retrieve | Reads the results its crawler can reach | Your robots.txt blocks OAI-SearchBot |
| 3. Compare | Looks for brands that several pages agree on | Only your own site mentions you |
| 4. Write | Names a short list with a reason for each | Nothing says what you are best for |
Even when ChatGPT answers without searching, it leans on what it learned in training, which is the same web at an earlier date. Either way the raw material is text about your brand, and my breakdown of how ChatGPT chooses which sites to cite follows that path one stage at a time.
The Sources ChatGPT Reads Most
The sources ChatGPT reads most are Wikipedia, Reddit, and large publishers. Profound analyzed 680 million AI citations and found Wikipedia held 47.9% of ChatGPT’s top ten sources, with Reddit at 11.3% and Forbes at 6.8%.
Those three stand for three kinds of sites, and each one does a different job for a brand:
| Layer | Examples | What it tells ChatGPT | What a store can do |
|---|---|---|---|
| Reference | Wikipedia, directories | The basic facts about your brand | Keep one consistent description everywhere |
| Review | Reddit, review sites | What buyers say after they purchase | Answer threads and ask customers for reviews |
| Editorial | Forbes, niche blogs | Which brands belong on a short list | Pitch roundups and comparisons |
But that mix moves without warning. Semrush tracked over 100 million AI citations across 13 weeks and watched Reddit fall from about 60% of ChatGPT responses to about 10% in September 2025. Wikipedia dropped from about 55% to under 20% over the same stretch.
A brand that built its whole plan around one of those sites would have lost its footing in a change it never saw. So spread your mentions across all three layers, and one model update cannot erase you.
Of the three, the editorial layer is the one a store can move fastest. “Best of” roundups hand ChatGPT a short list of names for the exact questions shoppers ask, and I cover how to get on them in whether listicles get you cited by AI.
Why Do Mentions Beat Backlinks?
Mentions beat backlinks since ChatGPT reads text about your brand, not a link graph. Ahrefs studied 75,000 brands and found branded web mentions correlated with ChatGPT visibility at 0.664, while Domain Rating sat at 0.266. YouTube mentions scored highest at 0.737.
| Signal | Correlation with ChatGPT visibility |
|---|---|
| YouTube mentions | 0.737 |
| Branded web mentions | 0.664 |
| Branded anchor text | 0.511 |
| Branded search volume | 0.352 |
| Domain Rating | 0.266 |
Read from the top, that table is a ranking of how much the web talks about you. The classic authority score sits last. Most SEOs still price a backlink above a mention, and I think that order is backward for AI answers, which is the case I make in brand mentions vs backlinks.
Branded search volume sits in the middle of the table for a reason, too. People typing your name into a search bar is proof the brand exists outside its own website. A store that grows that number is building the same signal the model looks for.
Google rankings carry over less than you would expect. Semrush found about 90% of the pages ChatGPT cites rank in position 21 or lower in Google for related searches. A page-one ranking and a ChatGPT recommendation are two separate contests with a few shared rules.
Does your own website still matter?
Your own website still matters, mostly as the place ChatGPT checks facts. That same Semrush study found half of the links in ChatGPT responses point to business and service websites, so the model does visit brand sites for prices, specs, and policies. Mentions get you named, and your site confirms the details.
Do YouTube mentions count that much?
YouTube mentions had the highest correlation in the Ahrefs study, at 0.737. The model does not need to watch the video, since titles, descriptions, and transcripts are all text. A product named in ten review videos has ten more pages agreeing on what it is, so one reviewer with a small channel in your category is worth a pitch.
Product Recommendations Run on Feed Data
Product recommendations in ChatGPT run on feed data more than on articles. OpenAI’s shopping documentation says product results are “selected independently by ChatGPT and are not ads,” and that merchants are ordered by “availability, price, quality, and whether they are the maker or primary seller.”
For a Shopify store, the door is already open. That same page says Shopify product data reaches ChatGPT through Shopify Catalog, so you do not apply for anything. What decides the outcome is how complete your data is when it arrives, and I cover how to get ChatGPT to recommend your products step by step in its own guide.
So the fields do the selling. A title that names the product type, a current price, an accurate in-stock status, and a real review count give ChatGPT everything it sorts on. A product missing one of them loses to a competitor’s listing that has all four, even if your product is better.
Reviews deserve a second look while you are in there. A product with 4 reviews and a product with 400 can carry the same star rating, and the count is part of what a shopper sees in the answer. Turn on review requests after every order if you have not already.
That cleanup is where I start in my AI search services for Shopify stores, since it is the fastest fix on this page. Mentions take months to build, and a product feed can be corrected in a week.
The “maker or primary seller” line matters for brands that also sell through retailers. If Amazon or a big-box store carries your product, ChatGPT may show their listing next to yours, and the more complete one tends to win. Keep your own store’s listing the most detailed version that exists.
Nobody can buy the recommendation, which means somebody earns it.
Why Is ChatGPT Recommending Your Competitor?
ChatGPT is recommending your competitor for one of four reasons, and the most common one is that it can read their store and not yours. The other three are mentions, clarity, and product data. Check them in that order, since each one blocks the ones after it.
1. Your robots.txt blocks OAI-SearchBot
Open yourstore.com/robots.txt and search for “OAI-SearchBot” and “GPTBot.” Some apps and security tools add a block for every AI crawler without asking. OpenAI says a change to that file takes about 24 hours to register, so this is a one-day fix.
2. Only your own site describes you
Search your brand name with the word “review” and count the results that are not yours. If the first page is your store, your social profiles, and nothing else, ChatGPT has no second opinion to confirm you, and I cover where it looks in do AI models read your reviews. Your competitor likely has three roundups and a Reddit thread doing that job.
The fix is outreach, and it can start small. Send your product to three reviewers in your category, and ask two happy customers to post about it where they already spend time online.
3. Your pages never say who the product is for
ChatGPT recommends a brand for a need, like “best for small apartments” or “best under $200.” If your product page lists features and never names the shopper it fits, the model has to guess. That guessing is also why ChatGPT makes things up about brands. Add one sentence near the top of each product page that says who it is for and why.
The same goes for collection pages. “Standing desks for small apartments” tells the model more than “Our Desk Collection” ever will.
4. Your product data is thin or out of date
A wrong price or a stale stock status can drop a product from a shopping answer, even when everything else is right. Compare your feed to your live product pages once a month. Fix any field that does not match.
Can a Small Brand Get Recommended Over a Big One?
A small brand can get recommended over a big one when the question is narrow. “Best running shoes” goes to the names every source repeats. “Best running belt for a large phone” has a handful of brands with any coverage at all, and a focused store can be the one most sources agree on. That is the whole case for why small brands can beat big brands in AI answers.
So pick five narrow questions where your product is the obvious answer, and build your mentions around those. One roundup, one Reddit thread, and one review video on the same narrow topic can be enough agreement to land on a short list of three. Win the narrow questions first, then widen.
Just know the wins come one question at a time. If someone tells you a single post got them recommended everywhere, be skeptical. Narrow answers come first, and the broader ones follow as the mentions stack up.
Become the Brand ChatGPT Recommends
You become the brand ChatGPT recommends by fixing access first, then data, then mentions. That order follows the path an answer takes. The crawler has to reach your store before the feed or the mentions can count for anything.
- Open the door. Allow OAI-SearchBot in robots.txt, and check that your key pages are indexed in Bing as well as Google.
- Complete the product data. Fill in titles, prices, availability, and reviews for every product in your Shopify catalog.
- Say who each product is for. Put the use case in the first two sentences of the product page and the collection page.
- Earn mentions on other sites. Pitch the roundups, reviewers, and YouTube channels in your category, and answer questions on Reddit under your own name.
- Keep the story consistent. Use the same one-line description of your brand on your site, your profiles, and every directory listing.
Each of those fixes runs on its own clock. OpenAI says a robots.txt change takes about 24 hours to register, while mentions have to be published, indexed, and read before they count:
| Fix | Time to do | Time to show up |
|---|---|---|
| Crawler access | Ten minutes | About a day |
| Product data | About a week | A few weeks |
| Mentions on other sites | Ongoing | A quarter or more |
Once all five are in place, give them a full quarter before you judge the result. Mentions build slowly, and the brands being recommended today mostly did this work months ago. The content side of the same plan is in my 7 steps to rank on ChatGPT.
The payoff is worth the wait. Semrush found the average AI search visitor is worth 4.4 times the average organic visitor by conversion rate, since the answer has already narrowed the choice before the click.
While you wait, track brand mentions on ChatGPT the simple way. Ask ChatGPT your own buying questions once a month and write down who it names. A brand that shows up in two answers out of ten this month and four next month is on the right track, whatever a dashboard says. Those mentions pay off, since AI citations convert into customers at a higher rate than most traffic.
See What ChatGPT Says About Your Brand
ChatGPT recommends the brands its crawler can reach, its sources keep mentioning, and its product data can confirm on price, stock, and reviews. Backlinks and Google rankings matter less than how widely the web talks about you, and no budget buys a placement.
If you want to know why the model names your competitors and not you, book a call. I’ll run your top buying questions through ChatGPT and show you which of the four gaps is holding you back, and my AI search services cover the fixes if you want them handled.
Sources:
- OpenAI: Overview of OpenAI crawlers
- Seer Interactive: 87% of SearchGPT citations match Bing’s top results
- Profound: AI platform citation patterns
- Semrush: The most-cited domains in AI
- Ahrefs: Top brand visibility factors in ChatGPT, AI Mode, and AI Overviews (75,000 brands)
- Semrush: AI search and SEO traffic study
- OpenAI Help Center: Shopping with ChatGPT search
