Reading time: 13 minutes
To get ChatGPT to recommend your business instead of a competitor, your brand must appear in the sources it reads for category questions. These include lists, reviews, comparisons and press. In general, a competitor gets recommended by ChatGPT because those sources talk about it more. This guide shows how to find that gap and get AI to recommend your business.
In short
- In a Genezio study, when the user asked for a recommendation, 16.5% of ChatGPT’s internal searches already included a competitor the user had not named.
- In category questions, only 7.2% of brand mentions came with the brand’s own site cited, according to BuzzStream. In other words, the brand’s site is almost never the source of the pick.
- In an Ahrefs study of “best X” questions, blog lists made up 43.8% of the pages ChatGPT cited.
- No study has measured how long it takes to become the pick. With search, sources change every week; without search, change depends on a new model.
Why does ChatGPT always recommend the same brands?
ChatGPT recommends the same brands because it picks from the ones it already links to the category, in training and in the pages it finds. The list changes with each answer, but this pool of candidates repeats. So when a competitor is recommended by ChatGPT, it is because it sits in that pool, and your brand is outside or at the edge.
SparkToro tested this with Gumshoe.ai, which sells AI tracking. In total, 600 volunteers ran 12 questions 2,961 times in ChatGPT, Claude and Google’s AI. The chance of ChatGPT repeating the same list of brands was below 1 in 100. Still, some brands showed up almost every time. For example, in a test with 142 different questions about headphones, the most cited brands appeared in 55% to 77% of 994 answers. Rand Fishkin, cofounder of SparkToro, sums up what these lists are:
“These are not endorsements. They’re not ordered or chosen by any set of criteria other than the spicy autocomplete of AI modeling: tokens that follow other tokens frequently in the corpus (and/or in the extracted results when RAG is used).”
Rand Fishkin, SparkToro (2026)
Also, being in that pool goes hand in hand with what the web says about the brand. For example, Ahrefs analyzed 75,000 brands in ChatGPT, AI Mode and AI Overviews. Brand mentions on the web had a correlation of 0.66 to 0.71 with presence in these AI tools. YouTube mentions were even stronger, at about 0.737. However, Ahrefs sells SEO tools and notes that correlation is not causation. Changing this pool of candidates is the goal of the work described in the guide to what GEO is.
ChatGPT adds the competitor to its own search
With search, ChatGPT turns the question into queries before it reads any page, as the guide on how to show up in ChatGPT explains. Genezio, which sells an AI visibility platform, analyzed 220,000 of these internal queries. In 9.9% of them, ChatGPT added a competitor the user had not named. When the request was for a recommendation, however, the rate rose to 16.5%.
In addition, 97% of the brands that entered these queries made it to the final answer. In other words, ChatGPT recommends a competitor it already knows and searches for it. Your brand, on the other hand, must show up in the pages the search finds. However, the study only observed ChatGPT and does not say which country or language it covered.
What is the difference between being cited and being recommended?
Being cited means your site is used as a source for the answer, with a link. Being recommended means your name is suggested to the customer as an option. The two are separate, because the AI can cite your page and recommend the competitor. So, for ChatGPT or any other AI to recommend your business, your site alone is not enough.
BuzzStream, which sells PR software, analyzed 12,000 answers from four AI engines by Google and OpenAI. Overall, only 23.1% of brand mentions came with the brand’s site cited in the same answer. In category questions, such as “best project management tools,” the rate fell to 7.2%. By contrast, it was 35.8% in head-to-head comparisons of two brands. However, the study does not say which country the questions came from.
A list where the brand ranks itself first shows the same effect. For example, in a 2026 review of B2B software searches by SEO consultant Lily Ray, Google’s AI used these lists as sources. But it left the publishing brand out of the recommendation in most cases. Therefore, track citations and recommendations in separate columns. After all, when your page is cited but a competitor is recommended by ChatGPT, you won the citation and lost the sale.
What sources does the competitor have that your brand lacks?
In general, the competitor recommended by ChatGPT appears in more third-party sources that the AI reads in the category. These are “best X” lists, review sites, comparisons, press, videos and communities. Its own site, however, matters less than it seems. Thus, to get ChatGPT and other AI to recommend your business, the gap to close is usually outside your site.
| Source | What counts | How to check |
|---|---|---|
| “Best X” lists | Presence and position | How many cited lists include each brand |
| Review sites | Number and date of ratings | Recent reviews of each brand |
| Comparisons | Who publishes and who wins | Cited “X vs Y” and alternatives pages |
| Press | Stories about the category | News stories that mention each brand |
| Videos | Tests and reviews | Videos that mention each brand |
| Communities | Customer stories | Threads on Reddit and industry forums |
What do lists and reviews show?
Lists show up a lot among the sources. Ahrefs analyzed ChatGPT’s answers to 750 “best X” questions about software, products and agencies. Blog lists made up 43.8% of the cited pages. Also, the brands ranked higher in third-party lists appeared more in the recommendations, but this is correlation, not causation.
Reviews and comparisons also enter the search. In the Genezio study, ChatGPT added “reviews” to queries in almost every industry. Reviews and comparisons each appeared in 18% of queries, and pricing in 10%. Each industry, however, has its own authorities. For example, in sales and B2B software, the study names G2 and Gartner.
In Brazil, for example, the list also includes Reclame Aqui (a Brazilian complaints site), Google reviews and marketplaces. The rules for gaining presence in each source are in the guide to the sources most cited by AI. Also, if the AI describes your brand with a wrong fact, such as an old price or a closed location, the competitor wins by default. In that case, see what to do when ChatGPT is making up information.
How do you find the questions where ChatGPT recommends a competitor?
Build a list of questions without your brand and run each one several times. In each answer, write down the brands recommended and the sources cited. Also include questions with the competitor’s name. Then compare the sources: the ones that cite the competitor and not you form your target list.
- Category and problem: write the questions the way customers ask them, with city, company size and price range when it makes sense.
- Competitor: “alternatives to [competitor],” “is [competitor] worth it?” and “[competitor] or [your brand].” In a G2 survey of 1,076 software buying decision makers, 31% start their AI research with a question about a competitor.
- Repetition: run each question several times, in a temporary chat, and mark the brands recommended by ChatGPT. Free and paid checkers are covered in our guide to the AI visibility checker.
- Sources: open the sources of each answer and note whether they cite the competitor, your brand or both.
- Piles: sort the sources into third-party content, competitor pages and context.
How do you sort the competitor’s sources?
The split into piles comes from Garrett French, founder of Citation Labs, in a comment that BuzzStream published in the study itself. For him, third-party lists, reviews and comparisons become outreach targets, while Wikipedia and Reddit serve as context. So each source in the first pile is a chance for AI to recommend your business. About the competitor’s pages, he says:
“Study competitor pages getting cited instead of yours. They may be doing something your similar pages aren’t, such as comparison detail, implementation specifics, or constraint-aware guidance.”
Garrett French, Citation Labs (2026)
On the other hand, when the answer shows no source, it comes from the model’s training. In that case, ChatGPT recommends a competitor based on what the web said about it before the cutoff date. So the path is the same, only slower.
How do you explain to a client why AI recommends a competitor?
Show the source map, not a single screenshot. After all, a screenshot shows one answer, which changes in the next one. The map, by contrast, shows in how many answers the competitor appears. It also shows which pages support the competitor recommended by ChatGPT and which of them do not cite the client. Finally, show what is missing for ChatGPT to recommend the client’s business in those sources. That way, the talk moves from complaint to a plan with targets, owners and a review date.
What should you do first to get ChatGPT to recommend your business?
Start with what takes your brand out of the race, then move to the sources the AI cites most in your category. The practical order is this: fix the basics, get into lists and reviews, publish honest comparisons and give the press a reason to cover the brand. In short, this is how you get ChatGPT to recommend your business.
- Fix the basics. If your brand does not show up even when someone asks for it by name, the problem comes before the competitor. See the 8 reasons AI does not cite a brand and how to check each one.
- Get into the cited lists. Give authors facts they can check, such as pricing, case studies and trial access. Before publishing a list of your own, see when a self-promotional list helps or hurts.
- Ask every customer for reviews, with no incentive, and reply to criticism in public.
- Publish the pages the search looks for: category, pricing, “X vs Y” and alternatives. An honest competitor comparison page shows criteria, pricing and who each option is for.
- Give the press a reason: your own data, studies and customer cases.
Why does the comparison page still count?
The comparison page matters even when the customer already names both brands. In the Genezio study, when the user asked to compare two brands, ChatGPT searched for only one of them in 25% of conversations. In half of those cases, the tracked brand was left out. Therefore, without pricing and comparison pages, it is harder to get ChatGPT to recommend your business in a direct comparison.
What does not help AI recommend your business?
A large number of pages does not replace third-party sources. In the Ahrefs study of 75,000 brands, the number of pages on a site had almost no link to presence in AI tools. The correlation was about 0.194. In addition, Google’s guide says that seeking inauthentic mentions is not as helpful as it might seem. So bought reviews, paid lists without disclosure and mass pages bring risk and little gain for getting AI to recommend your business.
How long does it take to get recommended by ChatGPT?
We found no public study that measures how long a brand takes to become the pick. After all, the timing depends on the route. With search, a new source can enter the answer soon after it is indexed. Without search, however, change only comes when a model trained on newer data goes live.
ChatGPT’s sources turn over fast. SISTRIX, which sells SEO tools, tracked 82,619 questions for 17 weeks in five European countries and the US. In ChatGPT, each answer cited 3 to 4 domains on average. But the weekly turnover of these domains is only given for three countries: 74% in Germany, 60% in the UK and 42% in France. In other words, a new list can enter quickly, but it can also drop out the next week.
Likewise, recent lists have an edge. In the Ahrefs study, 26% of the 1,100 lists ChatGPT used as sources had been updated in the previous two months. That is why a brand that wants to stay recommended by ChatGPT needs to be in many sources, not just one.
Measure every month, with the same questions and the same competitors. Then compare the presence rate, not the position on the list. That way, when a competitor is recommended by ChatGPT in fewer answers and your brand in more, the change shows up first in that rate. That is the proof that you got ChatGPT and other AI tools to recommend your business.
Frequently asked questions
No. According to OpenAI, ads in ChatGPT may appear below the answer, labeled as sponsored. Also, advertisers cannot influence or change what the answer says. So whether ChatGPT recommends a competitor or your business depends on what it finds in training and in the sources.
Often, yes. In the Ahrefs study of 75,000 brands, the brands mentioned by ChatGPT, AI Mode and AI Overviews had a correlation of 0.749 to 0.821 with each other. But sources change from one engine to another, so the source map needs to be built for each AI.
Yes, but the sources change. For a local business, check Google reviews, city directories and the local press. Also, to see whether ChatGPT or another AI will recommend your business in a city, include the neighborhood in the questions. Then note which profiles and pages the answer cites.
Want to know how your brand shows up in AI answers today?
NotHuman is opening spots for agencies and brands to track how they appear in AI answers. Join the waitlist and we’ll let you know when a spot opens up for you.
Main sources
- SparkToro, study on how inconsistent AI brand recommendations are (2026).
- Genezio, how ChatGPT picks competitors in its internal search queries (2026).
- BuzzStream, study on brand mentions and citations in AI answers (2026).
- Ahrefs, visibility factors for 75,000 brands in ChatGPT, AI Mode and AI Overviews (2025).
- Ahrefs, study of 26,283 sources cited by ChatGPT in “best X” questions (2025).
- SISTRIX, stability of sources cited in AI search results (2026).
- G2, survey of software buyers on AI search (2026).
