ChatGPT – how your company shows up in the answers, and why your competitor gets named instead

Definition

ChatGPT is OpenAI's AI assistant. It answers questions, writes text and code, and searches the web. As one of the most widely used AI services in the world, it is also a channel where companies are found, compared and described – either from the model's memory or from a real-time web search.

What is ChatGPT?

ChatGPT is an AI assistant from OpenAI, launched in November 2022. You ask a question in plain language and get a coherent answer back: explanations, drafts, code, analysis and research. The service is available as a web app, a mobile app and an API for developers. ChatGPT passed 900 million weekly active users in February 2026, according to TechCrunch. That makes it one of the places where potential customers now ask their first question about a supplier.

How does my company show up in ChatGPT's answers?

Your company shows up in ChatGPT when the model either remembers you from its training data or finds you in a real-time web search. Attached files and connected apps are exceptions, but when a buyer asks about suppliers, these are the two routes that matter. Either way, the raw material is whatever can be read about you – on your own website and on other people's.

The two routes work differently. From memory, ChatGPT answers with whatever was in the training data when the model was built. That data is often months old, and for small and medium-sized companies it is thin or missing altogether. With web search (ChatGPT Search), the model instead retrieves current pages, picks out the passages that answer the question and cites them with links. That is the route a company can influence in the short term.

Three kinds of material decide what ChatGPT retrieves. First, pages that answer the question clearly: a direct answer under a heading that mirrors the question, in a passage that can be lifted out on its own – what is known as semantic chunking. Second, what third parties write about you: industry lists, comparisons, reviews and ratings carry a lot of weight when the question is "who is good at X". Third, consistent information about who you are: the same company name, services, address and contact details on your website and in your profiles elsewhere – your Google Business Profile, Bing Places, LinkedIn and industry directories – ideally backed up with schema markup. If the profiles disagree, the model picks one of them, and not necessarily the current one.

You cannot buy a place in the answer itself. Since February 2026, OpenAI has shown ads to logged-in users on the Free plan and the low-cost Go plan in a growing number of countries (OpenAI Help Center). They appear below the answer, are labelled as sponsored and, according to OpenAI, do not affect which companies the model mentions when someone asks for a recommendation. An ad is not a citation.

Why does ChatGPT mention our competitor but not us?

ChatGPT mentions your competitor because there is more to read about them, and you can find out where the gap is. Three checks are enough for a first diagnosis.

1. Is the model answering from memory or from a search? Repeat the question and ask ChatGPT to list its sources. If it shows links, it has searched the web, and the question becomes why your pages were not chosen: is the answer on your website, and is it stated clearly enough?

If it shows no sources, it is answering from memory, and the information about you is either too thin or too recent to have stuck. When ChatGPT lacks information about you, it rarely answers "I don't know". It answers with what it has – often a competitor that has spelled out the answer. The difference between being named (mentions) and being used as a source (citations) is explained in the section Why does AI mention your competitor but not you?.

2. What do third parties say about you? AI answers are built on what others write about you – brand mentions – not just on your own website. Search for your company name along with your service and see what comes up: industry websites, directories, review sites, LinkedIn, podcasts. If your competitor appears on three "best agencies" lists and you appear on none, that is the gap ChatGPT reproduces. Ahrefs analysed 75,000 brands and found a correlation of 0.664 (Spearman) between web mentions and how often a brand is mentioned in AI Overviews. For backlinks, the figure was 0.218. The study covers Google's AI Overviews and established domains (DR above 40), not ChatGPT, but the mechanism is the same: the model trusts what several sources say.

3. What is missing or blocked on your side? Start with the content. ChatGPT can only reproduce what has been written down, so the first question is whether the answer is on your website: one page per question your buyers ask, with a direct answer at the top, headings phrased as questions and FAQ pairs for follow-up questions. If there is no page about a service, an integration or a pricing model, the model cannot retrieve it, however good the rest of your website is. We explain how to check your most important pages against that checklist in Phase 1: the basics on your own website, and how to build a text whose sections can be lifted out in How do you structure a text so that both Google and AI can read it?.

Then come three technical checks. First, robots.txt: according to OpenAI's documentation, it is OAI-SearchBot that determines whether a page can appear in ChatGPT's search results, while GPTBot collects material for training and ChatGPT-User fetches a page when a user asks for it. The settings are independent of each other. If you blocked GPTBot to stop training, check that OAI-SearchBot was not blocked at the same time. Second, make sure the text is in the HTML, not just rendered with JavaScript. There is nothing to indicate that OpenAI's crawlers run scripts, so put important text in the page source itself. Finally, does the page exist at all? A service that is only mentioned in a PDF or on a sales call cannot be retrieved.

If ChatGPT describes you wrongly rather than not at all – the wrong price, the wrong service, a founder who never existed – that is a different problem. It is covered in the glossary entry on AI hallucination.

Test it yourself: how you show up in ChatGPT

Test with ten fixed questions, the same ones every time, and keep a record of the answers. Open a new temporary chat for each question so that earlier answers do not colour the next one. Ask each question word for word. Note whether ChatGPT used web search (you will see source links) or answered from memory, along with the date and the model.

The questions below are the ones we use ourselves. Replace [service], [company], [city] and [country] with your own.

1. "Which providers in [city] are good at [service]?"
2. "Recommend a [service] provider for a B2B company in [country]"
3. "Does [company] do [specific service]?"
4. "Where is [company] based and what do they do?"
5. "Compare [company] with other [service] providers"
6. "How much does [service] cost in [country]?"
7. "Which companies in [country] offer [niche service]?"
8. "How do I choose a [service] provider?"
9. "Is there a provider that can handle [the requirement your customers usually raise]?"
10. "What do customers say about [company]?"

If you would rather not do it by hand, AEO-kollen, our own free Swedish tool, does the same job. It asks ChatGPT, Perplexity and Claude 10–15 industry questions, three times each, and reports whether you are mentioned, where, whether the description is accurate and which competitors are named instead. Both the questions and the raw answers are shown.

Each run is a snapshot. The same question can get a different answer the next day, naming different companies, so compare patterns across several runs rather than individual answers.

How do you measure visibility in ChatGPT over time?

You measure visibility in ChatGPT by asking the same fixed questions before and after a change and comparing the answers, backed up by three data sources that do not depend on the questions you ask. Bing Webmaster Tools provides first-party data for Copilot, showing which pages are cited and for which queries. Your server logs show whether OAI-SearchBot and ChatGPT-User actually fetch your pages, and how often. GA4 shows visits with chatgpt.com as the referrer – people who clicked through from an answer. The glossary entry on AI visibility brings together the method, the tools and what not to measure.

This is the method we use in our own AI SEO work (see AI SEO agency). We cannot yet show a verified figure for AI visibility: the answers vary too much for a percentage to be an honest measure. For Learnifier, where we handle the ongoing SEO and AEO work on a website in four languages, some deals have started in AI assistants.

What is ChatGPT Search and how does it affect SEO?

ChatGPT Search is ChatGPT's built-in web search: the model retrieves current pages, answers the question and lists its sources with links. That turns ChatGPT into an answer engine rather than just a chatbot, and takes over some of the questions people used to ask Google. It is OpenAI's own search bot, OAI-SearchBot, that determines whether a page can appear in ChatGPT's search results. It also does no harm to be indexed in Bing, which Microsoft's Copilot is built on.

For SEO, the goal broadens. Ranking is not enough; the page also has to be retrievable and citable, which is what AEO is about. The foundation is the same as for good SEO: technical accessibility, clear structure and content that answers the buyer's questions. We explain how ChatGPT, Perplexity and Google AI Overviews differ in AI search optimisation in practice.

Facts about ChatGPT: models, Codex, pricing and ownership

Models. ChatGPT runs on OpenAI's GPT models, a type of LLM trained on very large amounts of text. In November 2026, OpenAI is rolling out GPT-6 to paying users on Plus, Pro, Business and Enterprise (OpenAI Help Center); free users get an older model by default. The models are replaced several times a year, so the version you get depends on your account and plan. The default model is enough for most everyday work. Reasoning variants are better suited to multi-step tasks.

Codex. Codex is OpenAI's agentic coding system – an umbrella name for the terminal tool, IDE extension, cloud service and GitHub integration that let ChatGPT carry out development tasks on its own: planning, running and verifying work over several steps. We explain how AI-assisted development works in practice in the article on AI-assisted web development.

Pricing. The basic version is free and only requires an account. Paid subscriptions such as ChatGPT Plus give access to the latest models, higher usage limits and more features. For businesses there are Business and Enterprise, with shared workspaces and admin controls. Prices change regularly; see OpenAI's pricing page for current levels.

Ownership. ChatGPT is owned by OpenAI, an American AI company founded in San Francisco in 2015. Sam Altman is its CEO and a co-founder. Microsoft is the largest external shareholder, with a stake of around 27 per cent following the restructuring in October 2025 (Fortune). As well as ChatGPT, OpenAI develops image and video models and APIs that other companies build their services on.