Why your competitor shows up in ChatGPT and you do not
It is rarely because they are better, and almost never because they paid for it. Here is how an assistant decides whom to name, the six gaps that usually explain the difference, and a test you can run on your own business in ten minutes.
AI assistants name the businesses the rest of the web already describes clearly and often. If your competitor gets recommended and you do not, the usual reasons are that other websites mention them more, their listings and reviews are fuller and more consistent, and their pages can actually be read by AI crawlers. In a study of 75,000 brands, Ahrefs found that mentions of a brand across the web tracked AI visibility about three times more closely than backlinks did. Most of the gap is fixable. The quick parts take days, and the part that matters most takes months.
Is your competitor really showing up, or did you just see it once?
Check this before anything else, because AI answers move around far more than search results ever did.
In January 2026, Rand Fishkin of SparkToro and Patrick O'Donnell of Gumshoe had 600 volunteers run the same 12 prompts through ChatGPT, Claude and Google's AI nearly 3,000 times. The chance of getting the same list of brands twice was under 1 in 100. The chance of getting the same list in the same order was closer to 1 in 1,000.
What did hold steady was how often a brand appeared. Some brands turned up in 60 to 90 per cent of answers for a given need, whatever the exact wording of the question. Smaller, more specific markets were also more stable than big, crowded ones.
"Number one in ChatGPT" is not a real position. "Named in seven answers out of ten" is.
So one screenshot from a client or a friend proves very little either way. Run the question several times, on more than one assistant, before you decide there is a problem. The test further down is built around exactly that.
How does an assistant decide which businesses to name?
It draws on two sources, and your competitor may be winning on either one.
What the model already learned
Every assistant is trained on a large slice of the public web, frozen at a point in time. If a business was written about widely before that point, the model has an impression of it before anyone asks. A business that opened recently, or that nobody wrote about, simply is not in that memory.
What it looks up when you ask
For anything current or local, the assistant runs a web search, reads a handful of pages and summarises them. Here it can only recommend what it can find, and it leans heavily on a few kinds of source. Conductor tracked AI citations from September 2025 to March 2026 and found Wikipedia leading ChatGPT's citations for recommendation questions every month, Reddit leading its answers to support questions, and YouTube at or near the top for Perplexity, Gemini and Google's AI Overviews.
Notice what is missing from that list: your own website. It matters, but it is rarely the main thing an assistant reads when someone asks it to choose between businesses.
This is not a niche behaviour in India any more. In February 2026, OpenAI's chief executive Sam Altman said India had 100 million weekly ChatGPT users, its second largest market after the United States. Google's AI Mode launched here in 2025 and now answers in Hindi and seven other Indian languages. The questions your customers used to type into Google are increasingly being asked out loud, in full sentences, to something that replies with names.
What does your competitor have that you do not?
When we look at why one business gets named and a similar one does not, the answer almost always falls into one or more of these six gaps.
1. Other websites talk about them
This is the big one. Ahrefs studied 75,000 brands and found that the number of web pages mentioning a brand had a correlation of 0.664 with how often it appeared in Google's AI Overviews. Backlinks, the measure SEO has obsessed over for twenty years, came in at 0.218. The brands in the top quarter for web mentions averaged 169 AI Overview appearances. The next quarter averaged 14.
Correlation is not proof, and Ahrefs says so. But the direction matches everything else we see: news coverage, directory listings, "best of" roundups, forum threads, YouTube reviews and industry associations all count, and none of them live on your website.
2. Their details say the same thing everywhere
An assistant builds its picture of a business by combining what different sources say. If your Google Business Profile calls you a "digital agency", your website says "brand consultancy", your Justdial listing has an old phone number and your LinkedIn page lists a different city, you are hard to summarise. A competitor whose name, category, location and one-line description match on every page is easy to summarise, so it gets summarised.
3. They have reviews, and recent ones
For local and service businesses, reviews are a large part of what an assistant reads. BrightLocal's 2026 survey of US consumers found that 45 per cent had used an AI tool to get a local business recommendation in the past year, up from 6 per cent the year before. It also found that 97 per cent of those people sometimes checked the recommendation against real reviews. Reviews feed the answer and then confirm it. That survey is American, but the habit it describes is not.
4. Their website can be read by machines
This one is invisible from the outside and surprisingly common. Vercel looked at AI crawler traffic across its network and found that ChatGPT's and Claude's crawlers do not run JavaScript. Many modern websites, especially those built as web apps, load their actual text through JavaScript. A person sees a full page. GPTBot sees an empty one.
A second version of the same problem is a site that blocks AI crawlers outright, either in its robots.txt file or through a security setting. OpenAI states plainly that sites which opt out of its search crawler, OAI-SearchBot, will not be shown in ChatGPT search answers. Plenty of businesses have that block switched on without knowing it.
We had this problem ourselves. Our own site was originally built in a way that assembled each page in the browser, and the AI crawlers saw almost nothing. We rebuilt it so that every page arrives as finished text.
5. Their pages answer the question being asked
When someone asks for "a pediatric dentist in Dwarka who is good with nervous children", the assistant goes looking for pages that say something close to that. A competitor with a page about children's dentistry, naming the area and describing how they handle anxious patients, gives it something to quote. A homepage that says "Quality dental care for the whole family" does not.
Specific beats broad, and a page that answers its question in the first paragraph beats one that builds up to it.
6. They have been written about for longer
Sometimes the honest answer is age. A business that has had press coverage, reviews and listings for ten years is in the model's training memory. One that opened last year is not, and no amount of work will change what a model learned before you existed. What you can change is what it finds when it searches, which is why the first five gaps matter more than this one.
How can you test this yourself in ten minutes?
You do not need a tool or an agency to find out where you stand. You need a notebook and a phone.
- Write down three questions a real customer would ask before choosing someone like you. Use their words, not yours. "Best CBSE school near Dwarka with good sports facilities", not "leading educational institution".
- Ask each question in ChatGPT, Gemini and Perplexity. Use a temporary or logged-out chat so your own history does not tilt the answer. If your customers use Google, try AI Mode too.
- Ask each question twice. Note who appears both times, not just who appears once.
- Then ask each assistant directly: "What do you know about [your business name]?" Being described wrongly is often worse than not being mentioned, and it tells you which source is feeding the mistake.
- Ask where the answer came from. Perplexity and ChatGPT search show their sources. Write down which websites keep coming up for your competitor. That is your list of places to be.
- Open yourdomain.com/robots.txt in a browser. If you see GPTBot or OAI-SearchBot next to the word "Disallow", you are blocking ChatGPT.
- Right-click your homepage and choose "View page source". Search the page for a sentence you can see on screen. If it is not in the source, AI crawlers probably cannot read it either.
- Search your business name on Google and count how many results on the first page are not your own website or social profiles. Do the same for your competitor.
Record it in a simple table like this one, and repeat it once a month. Over a few months, the pattern is more useful than any single result.
Swipe to see the full table
| Question | ChatGPT | Gemini | Perplexity | Named you? | Described you correctly? |
|---|---|---|---|---|---|
| Question 1 | Names given | Names given | Names given | 0 to 6 times | Yes, no, partly |
| Question 2 | Names given | Names given | Names given | 0 to 6 times | Yes, no, partly |
| Question 3 | Names given | Names given | Names given | 0 to 6 times | Yes, no, partly |
What should you fix first?
Start with what is quick and certain, then move to what is slow and important. In our experience the order below wastes the least money.
Swipe to see the full table
| Fix | What it involves | How soon it can show |
|---|---|---|
| Let the crawlers in | Remove AI crawler blocks; make sure page text is in the HTML, not loaded by script | Days to a few weeks |
| Make your details match | One name, category, address, phone and description on your site, Google Business Profile, directories and social pages | A few weeks |
| Answer real questions on your site | A page for each main service or audience, with the answer in the opening paragraph and a short FAQ | Weeks |
| Build reviews steadily | Ask every satisfied customer, every month, and reply to every review | Ongoing |
| Earn mentions elsewhere | Directories, local press, industry lists, partner pages, YouTube, honest participation in forums | Months |
The timings are our working estimates, not guarantees. The last row is the slowest and the most important, which is uncomfortable but consistent with the evidence. We set out the difference between on-page and off-page work in more detail in SEO, AEO and GEO: what is actually different.
What we would not do
Three things, because this field is full of shortcuts that cost more than they save.
We would not pay anyone who promises to "rank you number one in ChatGPT". As the SparkToro study shows, there is no stable number one to rank for. Ask instead how they will measure how often you are named, across how many runs.
We would not buy reviews or post fake ones. They breach Google's policies, they get removed, and a sudden cluster of identical five-star reviews is easy for customers and platforms alike to spot.
And we would not flood Reddit, Quora or comment sections with self-promotion. Those communities remove it quickly, and a thread where people call a business out for spam is exactly the kind of page an assistant may later quote about you.
- AI answers vary from run to run. Measure how often you are named, not where.
- Assistants recommend what the rest of the web says about a business, so mentions elsewhere matter more than anything on your own site.
- Check the cheap things first: crawler access, readable pages and consistent details.
- Run the ten-minute test monthly. The trend tells you whether the work is landing.
Common questions
Why is my business not showing up in ChatGPT?
Usually because the web does not describe it clearly or often enough. ChatGPT recommends businesses that are mentioned on other websites, have consistent listings and reviews, and have pages its crawlers can read. A site that blocks AI crawlers, or loads its text through JavaScript, can be invisible to ChatGPT even when it looks fine to a person.
Can I pay to be recommended by ChatGPT or Gemini?
Not in the way people usually mean. Recommendations inside the answer are not sold as a ranking position, and studies show the lists change from one run to the next. What you can do is improve how often you are named by fixing your listings, reviews, website and mentions on other sites.
How long does it take to start appearing in AI answers?
Fixes to crawler access and to your own pages can show within weeks, because assistants search the live web for current questions. Being mentioned more widely elsewhere, which matters most, usually takes several months of steady work.
Does my Google Business Profile affect AI recommendations?
Yes, especially for local questions. Google's own AI features draw on Business Profile information, and other assistants read the same reviews and directory details. A complete, accurate profile with a steady flow of recent reviews is one of the most useful things a local business can maintain.
How do I check whether AI crawlers can read my website?
Open your robots.txt file to see whether GPTBot, OAI-SearchBot or similar crawlers are blocked. Then view your page source and search for text you can see on the page. If the text is missing from the source, it is being loaded by JavaScript, which most AI crawlers do not run.
Where these figures came from
The consistency findings are from the SparkToro and Gumshoe study, reported by Search Engine Land in January 2026. The citation source patterns are from Conductor's seven month analysis of AI citations. The brand mention correlations are from Ahrefs' study of 75,000 brands. The figure of 100 million weekly ChatGPT users in India is Sam Altman's, as reported by TechCrunch, and the AI Mode language expansion is also from TechCrunch. Consumer use of AI for local recommendations is from BrightLocal's Local Consumer Review Survey 2026, which surveyed 1,002 US adults. The finding that AI crawlers do not run JavaScript is from Vercel's analysis of AI crawler traffic, and the OAI-SearchBot statement is from OpenAI's crawler documentation. The six gaps, the fix order and the timings are our own working judgement. Most measurement in this field comes from companies that sell AI search services, and so do we, so treat the direction as reliable and the exact numbers as indicative.
Six years in content, creative direction and digital execution, most of it spent on the practical end of search, watching which pages actually earn attention and which only look busy. More about Megnon →