How to Get Your Company Found in ChatGPT

How does a company get found in ChatGPT?

ChatGPT names your company when it comes across your page during a search and recognises it as an answer to the question asked. Three things have to be right for that: First, OpenAI’s search crawler (OAI-SearchBot) is allowed to read your website. Second, your pages pick up your customers’ questions word for word and answer them in the first paragraph instead of writing around the topic. Third, the answer is backed up — with sources, numbers and concrete examples a model can carry over. Mentions on third-party comparison, directory and industry pages reinforce this, because AI search engines like to draw on several independent pages. Money only buys an ad next to the answer, not a mention inside it.

Last updated: 7 October 2026.

→ Comprehensive overview: This article covers the concrete actions. For the full, canonical guide on the topic, see AI Search Visibility: How B2B Websites Get Cited in ChatGPT, Perplexity, and More.

Why being cited matters

Where an AI answer appears, people click less. Pew Research Center analysed the browsing behaviour of 900 US adults in March 2025: users who saw an AI summary on Google clicked a traditional search result in 8% of visits — without a summary it was 15%. Users clicked a link inside the summary itself in only 1% of visits.^[1]

For a B2B company, that means a good ranking below the answer is worth less than it used to be. What counts is appearing inside the answer — as a named provider or as a cited source.

How ChatGPT, Claude and Perplexity find sources

AI assistants have two ways of knowing something about your company:

1. Training data

Whatever was publicly online before the training cut-off may have gone into the model. For existing models, you no longer have any influence on that. And for questions about providers, prices or current comparisons, training knowledge rarely suffices — so the model searches.

2. Live search

For current or purchase-related questions, assistants search the web and cite the pages they read. Providers run their own search crawlers for this, separate from the crawlers used for model training:

Provider	Crawler for search results	Crawler for model training	Fetch on user request
OpenAI (ChatGPT)	OAI-SearchBot	GPTBot	ChatGPT-User
Anthropic (Claude)	Claude-SearchBot	ClaudeBot	Claude-User
Perplexity	PerplexityBot	—	Perplexity-User
Google (AI Overviews, AI Mode)	Googlebot (regular search)	Google-Extended (control token, not a separate crawler)	—

Sources: OpenAI,^[2] Anthropic,^[3] Perplexity,^[4] Google.^[5]

The key consequence: if you block the search crawler, you won’t show up in that assistant’s search. OpenAI explicitly recommends allowing OAI-SearchBot so a website can appear in ChatGPT’s search results;^[2] Perplexity says the same for PerplexityBot.^[4] Whether you allow the training crawlers is a separate decision.

Google is simpler: for AI Overviews and AI Mode, a page only needs to be indexed and eligible to be shown in Google Search with a snippet. Google states explicitly that no special optimisation, no AI text files and no special schema markup are required.^[5]

What demonstrably helps in AI answers

The research paper that coined the term “Generative Engine Optimization” (Aggarwal et al., KDD 2024) tested nine methods on a benchmark of 10,000 queries. The result: the best methods — citing sources, adding quotations, adding statistics — raised visibility in generated answers by a relative 30–40%. Keyword stuffing, by contrast, did little.^[6] The effect varies by domain, but the direction is clear: evidence beats keywords.

Checklist: 10 steps to get found in ChatGPT

1. Allow the search crawlers in robots.txt

Check yourwebsite.com/robots.txt. If it contains Disallow: / for one of the search crawlers, you are invisible to that assistant. Allowing them looks like this:

User-agent: OAI-SearchBot
Allow: /

User-agent: Claude-SearchBot
Allow: /

User-agent: PerplexityBot
Allow: /

Also check whether your firewall or CDN blocks bots across the board — the providers publish their IP ranges for this.^[2][4]

2. Be properly indexed in Google and Bing

What isn’t indexed can’t appear in AI Overviews either.^[5] Search Console and Bing Webmaster Tools show which pages are missing. It’s classic SEO hygiene, but it’s the foundation for everything else.

3. Your customers’ questions as headings

Collect the questions prospects ask in first calls and turn them into headings — ideally word for word: “How much does … cost?”, “Is … allowed?”, “Who offers …?”. AI assistants are looking for an answer to exactly that question.

4. The answer in the first paragraph

The most common mistake: the page talks around the topic for three paragraphs before it answers. Put the answer directly below the heading, in two to four sentences that make sense without the rest of the page. Detail, reasoning and examples come after.

5. Provide evidence: sources, numbers, quotations

That’s exactly what had the strongest effect in the GEO study.^[6] Link studies, statutes and primary sources, give numbers with time frame and population, quote experts.

Bad (rarely cited):

We are a leading provider of innovative solutions for mid-sized companies.

Good (gets cited):

CegTec generated 41 SQLs for ProSeller AG in 6 months, at a 9.9% reply rate across 3 DACH segments.

Important: only numbers you can back up. A made-up number that an AI carries forward will come back to you eventually.

6. Make freshness visible

An assistant can only tell whether a page is current from what’s on it. Maintain a visible last-updated date on the page and dateModified in the Article schema — and actually update the content, not just the date.

7. Structured data where it adds clarity

Organization, Article and FAQPage markup make it unambiguous who you are, who is writing and which question is being answered. According to Google, though, markup is not a prerequisite for AI Overviews.^[5] Use it to add clarity, not as a shortcut.

8. Appear on third-party pages

AI answers to “Which providers are there for …?” often rely on comparison articles, directories and trade media. Check which pages are cited for your core questions and make sure you appear there correctly: industry directories, guest articles, comparison pages, review platforms.

9. llms.txt as an optional extra

llms.txt is a text file at yourwebsite.com/llms.txt that summarises your offering and most important pages in a format that’s easy for language models to read. It was proposed by Jeremy Howard in 2024 and is a proposal, not a binding standard.^[7] Google says it doesn’t need it.^[5] The effort is small, so it can be part of the mix — it is no substitute for steps 1 to 8.

# Company name
> One-sentence description.

## Services
- [Service 1](URL): Description
- [Service 2](URL): Description

## Contact
- Website: URL
- Meeting: Booking URL

10. Measure instead of guessing

Build a list of 20 to 50 questions your customers would ask an AI and check regularly in ChatGPT, Perplexity, Claude and Gemini: Are you named? Who is cited instead? What do those pages have that yours doesn’t — a direct answer, more evidence, a more recent date? That’s how we work at CegTec too: every gap is checked against the pages cited instead of us. A word of warning from practice: results fluctuate between two runs with identical questions. Read trends across several measurements, never a single one.

Can you buy your way into ChatGPT?

OpenAI has been showing ads in ChatGPT in the US since early 2026, and since late August 2026 also in 31 European markets including Germany and Austria — to users on the Free and Go plans; Plus, Pro and Enterprise stay ad-free.^[8] According to OpenAI, ads do not influence the answers.^[9] So paid visibility exists — but as a labelled ad next to the answer. You still only become a named provider or cited source inside the answer through the steps above.

The next stage: agents that work with you

AI agents don’t just read about your company; increasingly they can act — request meetings, query services, fill in forms. Offering a documented API (OpenAPI) or an MCP server makes it easy for agents to work with you rather than a competitor. For most B2B companies this isn’t the first step yet — steps 1 to 5 deliver more.

Conclusion

Getting found in ChatGPT is neither luck nor secret knowledge. Allow the search crawlers, answer your customers’ questions in the first paragraph, back up what you claim, keep pages current — and regularly measure who gets cited instead of you. If you’d like support with that: we run exactly this loop of measuring, diagnosing and improving for our own website.

Sources

1. Pew Research Center (22 July 2025): Google users are less likely to click on links when an AI summary appears in the results. Browsing data from 900 US adults, March 2025.
2. OpenAI: Overview of OpenAI Crawlers.
3. Anthropic: Does Anthropic crawl data from the web, and how can site owners block the crawler?
4. Perplexity: Perplexity Crawlers.
5. Google Search Central: AI features and your website.
6. Aggarwal, Murahari, Rajpurohit, Kalyan, Narasimhan, Deshpande: GEO: Generative Engine Optimization. KDD 2024.
7. Jeremy Howard: The /llms.txt file. Proposal of 3 September 2024.
8. OpenAI (18 August 2026): ChatGPT Ads expands across Europe.
9. OpenAI Help Center: Ads in ChatGPT.