Why doesn't ChatGPT recommend my brand? 7 measurable reasons

If ChatGPT does not recommend a brand, the reason is usually measurable: either you are measuring with the wrong questions, or ChatGPT cannot reach or read your site, or the pages it reads about you are on other sites and do not put you forward. This post covers seven reasons, each resting on a source or on one of our dated measurements, and how to check each one.

Going in order matters. The first reason concerns the measurement itself; looking at the others before fixing it may mean trying to solve a problem that does not exist.

How does ChatGPT find a brand?

ChatGPT does not look at the web for every question. According to OpenAI's help page, ChatGPT can search the web automatically when it judges that a question would benefit from up-to-date information; when it does search, it usually turns the user's question into one or more targeted search queries and sends them to search partners.

This means an answer can be built in two different ways: from sources found on the web, or from what the model learned beforehand. Our own measurement shows this difference clearly. Between 16 and 25 August 2026, on questions that did not name the brand for three corporate brands (a pharmaceutical company, a dairy brand, a technology group), 180 of ChatGPT's 362 measured answers (49.7%) cited at least one source. In the same measurement this was 339 of 339 answers on Perplexity. Unmeasured answers were not counted; since this is a three-brand measurement, the rate should not be read as general ChatGPT behaviour.

We cannot see from the outside what an answer with no cited source rests on. Some of the reasons below (access, JavaScript) affect answers where the web was searched; some (off-site sources, namesakes) affect both paths.

1. You are measuring with questions that contain your name

If you ask "X markası hakkında ne düşünüyorsun?" ("What do you think of brand X?"), X's name already appears in the answer. That is not a recommendation, it is an echo of the question. Whether ChatGPT recommends you is shown by questions a buyer asks before they know your brand, such as "Ankara'da kurumsal ekipler için hangi bordro yazılımını önerirsin?" ("Which payroll software would you recommend for corporate teams in Ankara?").

That is why the first step of diagnosis is to measure with Turkish questions that do not contain your brand name. This is how we measured our own product: on 27 August 2026 we asked 26 Turkish questions in the GEO category to five engines, and our product was not mentioned in any of the 116 measurable answers. Had we added questions containing our name, this zero would have turned into visibility that does not exist. The full method is in our guide on how to measure brand visibility.

Asking a single question once is not enough either. In research published by SparkToro in January 2026, when the same question was asked to ChatGPT 100 times, the chance of the same brand list coming back in two answers was under one percent. "I asked once and didn't come up" is not a diagnosis.

2. OAI-SearchBot cannot reach your site

According to OpenAI's crawler documentation, OAI-SearchBot is used to show sites in search results within ChatGPT's search features. The same document says sites that block this crawler will not be shown in ChatGPT's search answers, and may appear only as a navigational link. OpenAI's FAQ page for publishers also states that OAI-SearchBot must not be blocked for content to appear in summaries.

This crawler is separate from GPTBot, which is used for training. A common mistake is to block all OpenAI crawlers together in the robots.txt file while trying to block training. We explained which bot does what in our guide to robots.txt and AI bots.

The block may not be in robots.txt alone; it can also be a content delivery network or firewall setting. For example, on 1 July 2026 Cloudflare announced that it splits AI crawlers into three groups: search, agent and training. From 15 September 2026, on domains newly added to Cloudflare, agent and training crawlers are blocked by default on pages that show ads; the search group stays allowed by default. Check in your own settings which crawler falls into which group and what is blocked on your site.

3. Important information loads only with JavaScript

According to an analysis Vercel published on 17 December 2024 based on traffic on its own network, OpenAI's OAI-SearchBot, ChatGPT-User and GPTBot crawlers did not execute JavaScript. ChatGPT's crawlers downloaded JavaScript files but did not execute them.

If your product list, prices or service descriptions load with JavaScript after the page opens, ChatGPT's crawler may see an empty page. The check is simple: open the page's source HTML and see whether the important information is there as text. The analysis dates from 2024 and crawler behaviour may change; but serving important information in server-rendered HTML keeps you on the safe side in any case.

Next step

Is your site open to AI crawlers?

Enter your domain. The free scan measures whether AI crawlers can reach your site and whether your pages are quotable. No signup, no card, no email.

Your domainScan for free

4. The pages read about you are on other sites

When engines talk about a brand, they largely read pages outside the brand's own site. In our own measurement, across 60 analyses up to 3 August 2026, 93.6% of the 75,914 source links the engines showed were outside the brand's own site. The measurement covers ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews and rests on our own client and comparison brands, made up of 19 company records; it is not a sample of the market.

In the same measurement, these were the sites also cited by ChatGPT that appeared in the most analyses:

We included in the table only sites we could verify ChatGPT also cited; the engine list is across analyses, and how often ChatGPT cited these sites was not counted separately. What is written about your brand on these sites can affect how ChatGPT describes you: unresolved complaints, an incomplete Wikipedia article, an outdated product description on a marketplace.

There is also published data on the importance of off-site mentions, but not for ChatGPT. In Ahrefs' analysis of 75 thousand brands, the strongest relationship with being mentioned in Google AI Overviews was shown by the number of pages on the web where the brand's name appears (0.664); the relationship with backlinks was much weaker (0.218). Ahrefs stresses that this is a correlation and does not prove causation.

5. A namesake company is collecting your mentions

Another company with the same name may be sharing what ChatGPT knows about your name. For example, the name "GoPeople" can point both to an HR product and to a courier company in Australia. The answer to "GoPeople nedir?" ("What is GoPeople?") may be shaped by which one is written about more on the web.

The namesake problem also corrupts measurement: your name appearing in an answer does not mean it is talking about you. When counting a mention, check whether the answer describes your sector, your product or your city. Do not count a mention you cannot verify in your favour. Stating your brand's full name, what it does and where it is clearly on your site may help the engine tell you apart from your namesake; but you can only see the effect by measuring.

6. You were read but not named

ChatGPT may read your site and pass on the information without mentioning your name. In our own measurement, across 60 analyses up to 3 August 2026, in 341 of the 2,748 answers (12.4%) that cited the brand's own site, the brand name did not appear in the text of the answer. We determined whether the name appeared with automatic name matching; namesakes and spelling variants may affect the result.

This rate varied a lot by engine, and ChatGPT was in the lowest group:

So this reason exists on ChatGPT, but it is small compared with the other engines. If you see answers where your site is cited on ChatGPT but your name does not appear, check whether your pages present information as "we" or with the brand's name. Writing the name explicitly in the paragraph that carries the information may make it easier for the information's owner to be carried into the answer; we explained the writing rules for this in our guide to quotable content.

7. Your question set does not reflect the buyer's question

If the questions you measure are not the questions your buyer asks ChatGPT, the result does not describe your market either. If an enterprise software company measures with "en iyi yazılım hangisi?" ("which is the best software?"), it cannot see where it stands on the question the buyer actually asks: "100 kişilik bir ekip için bordro ve izin takibini birlikte yapan yazılım" ("software that handles both payroll and leave tracking for a 100-person team").

How a question is worded also changes what ChatGPT searches for on the web. OpenAI's help page says ChatGPT turns your question into targeted search queries; a question worded differently may lead to different queries and different sources. So questions should be written in the buyer's language, in Turkish and close to the purchase decision; and once set, they should be kept fixed so the next measurement can be compared.

Which reason should you start with?

Two of the reasons (2 and 3) are technical and can be checked in a few minutes. Our free scan shows, when you enter your domain, whether AI crawlers can reach your site and whether your pages can be read; it requires no sign-up. What the scan cannot tell you is whether ChatGPT actually mentions you: for that, real questions have to be asked and the answers read.

The other five reasons need measurement: questions that do not contain your name, separate results engine by engine, filtering out namesakes, and repeating with the same questions. After making a change, saying "it's fixed" without measuring again with the same questions is as much a guess as saying "there's a problem" without measuring.

Frequently asked questions

Is allowing OAI-SearchBot enough to appear on ChatGPT?

It is not enough, but it is necessary. According to OpenAI, sites that block this crawler are not shown in ChatGPT's search answers. Once you allow it, whether the pages are readable, what is written about you off your site, and how the question is asked also determine the result.

If I block GPTBot, will I become invisible on ChatGPT?

OpenAI describes GPTBot, used to collect content for training, and OAI-SearchBot, used to show sites in ChatGPT search, as separate crawlers. Blocking only GPTBot does not shut off OAI-SearchBot's access. Make sure your robots.txt file has separate rules for the two.

ChatGPT recommends me one day and not the next; why?

Answers to the same question can change. In SparkToro's January 2026 research, the chance of ChatGPT giving the same brand list in two answers to the same question was under one percent. So you need to look not at a single answer but at how often you are mentioned across many questions.

Is there a method that guarantees ChatGPT will recommend me?

No. Fixing the reasons in this post removes obstacles to ChatGPT finding and reading you; it does not guarantee a recommendation. The only way to see the effect is to measure with the same questions before and after the change.