How to Get Your Business Cited by ChatGPT and Perplexity

To get your business cited by ChatGPT and Perplexity, two things have to be true. Their search crawlers have to be able to reach your pages, and your pages have to answer the question better than the other sources they found. The first is a technical check most sites have never done. The second is ordinary good content. Neither involves paying anyone, because neither company offers a way to buy or request an organic citation.

This article covers ChatGPT search, Perplexity and, briefly, Claude. Google's AI Overviews work differently and have their own article; the AI search optimisation guide ties them together.

How ChatGPT and Perplexity find sources

When someone asks ChatGPT a question that needs current information, it searches the web, reads some results and writes an answer with citations. Perplexity works the same way by default. So the question is not "does the model know about my business" — it is "can the search step find my page".

Each company runs separate bots for separate jobs. This is the most important thing in this article, because the most common mistake is blocking all of them with one rule.

Sources, at the time of writing (October 2026): OpenAI's crawler overview, Perplexity's crawler documentation and Anthropic's crawler policy.

OpenAI states plainly that sites opted out of OAI-SearchBot will not be shown in ChatGPT search answers, and that robots.txt changes take about 24 hours to take effect. Perplexity says PerplexityBot is not used to train models and that sites should allow it to appear in Perplexity's results.

Step 1: check your robots.txt

Open yoursite.co.uk/robots.txt in a browser and read it. You are looking for any group that starts with User-agent: followed by one of the search bots above — or a wildcard — and contains Disallow: /.

The decision most businesses want is:

• Allow OAI-SearchBot, PerplexityBot and Claude-SearchBot, because these are what make you citable.
• Decide separately about GPTBot and ClaudeBot. Blocking them is a reasonable choice if you do not want your content used for training, and it does not affect search visibility in those products.

A pattern I see often is a block copied from a 2023 blog post that lists a dozen AI user agents under one Disallow rule. Some of those bots did not exist when the list was written; some that matter now are missing. Rewrite it deliberately, one decision per bot.

If you use WordPress, check whether your SEO plugin or a security plugin is generating robots.txt rules for AI bots on your behalf.

Here is how the common robots.txt patterns play out:

The last row sounds absurd, but a staging-site robots.txt that survives launch does exactly this, and it removes the site from ordinary search too. It is worth checking even if you have never thought about AI bots.

Note that the user-triggered fetchers sit outside this table: OpenAI says robots.txt rules may not apply to ChatGPT-User, and Perplexity says Perplexity-User generally ignores them, because a person asked for the page.

Step 2: check your firewall and CDN

Robots.txt allowing a bot does nothing if the server refuses it. Cloudflare, managed WordPress hosts and security plugins can all block or challenge AI crawlers, sometimes through a single toggle someone switched on months ago.

Check:

1. Your CDN or firewall dashboard for any setting that blocks AI bots or "AI scrapers".
2. Bot-fight or challenge modes that serve a JavaScript challenge to non-browser visitors.
3. Your server logs, if you can access them, for requests from OAI-SearchBot or PerplexityBot and the status codes they received.

OpenAI and Perplexity both publish IP ranges so you can allow their bots specifically rather than opening the door to everything claiming to be them.

Step 3: be indexed in Bing as well as Google

OpenAI says ChatGPT search uses its own crawler alongside third-party search providers, without naming them. Microsoft Copilot runs on Bing. Being properly indexed in Bing is therefore a sensible precaution, and many small sites have never checked it.

Set up Bing Webmaster Tools, submit your sitemap, and look at its index coverage. It also now has an AI Performance report, launched in public preview in February 2026, showing how often your pages are cited in Copilot and Bing's AI answers and the grounding queries used to find them. It is the closest thing to first-party citation data any AI search product currently offers.

Step 4: publish what is worth citing

Once the technical side is right, citation comes down to content. These are the patterns that make a page useful to an AI answer, drawing on how retrieval works and on the GEO research:

• Answer specific questions specifically. "How much does a boiler service cost in Kent" with a real price range and what affects it beats a page that says "competitive prices".
• Say things others have not. Original detail from your own work — what goes wrong most often, how long jobs really take, which option you recommend and why — is what makes you a better source than the next page.
• Back up facts. Link official sources for rules, fees and thresholds. It helps readers and it matches what performed well in the research.
• Keep facts about your business consistent. Name, address, services and prices should match across your site, Business Profile, directories and social profiles. Contradictions make you a less reliable source. Entity SEO and the Knowledge Graph covers this.
• Keep it current. Prices and rules that are two years out of date are a reason to cite someone else.

Before and after: a service page

Say a Medway removals firm has a page for house removals. It is a hypothetical, but a very common shape.

Before: a hero banner, "Stress-free removals across Kent", three paragraphs about the team's dedication, a gallery, and a contact form. Someone asking ChatGPT "how much does a 3-bed house move cost in Kent and how far ahead should I book" finds nothing on it to quote.

After:

1. An opening paragraph stating what the firm does, where, and the price range for typical moves, with what moves a job up or down that range.
2. A table by property size: typical crew, vehicle, time on the day, and price range — real figures from the firm's own pricing.
3. A section on how far ahead to book, based on the firm's actual diary pattern, including which months fill first.
4. A short list of what is and is not included: packing, dismantling furniture, insurance cover.
5. The firm's name, address and service area, matching its Business Profile exactly.

The second page answers the question a person actually asks, in a form that can be quoted, with information only that firm has. That is what makes a source worth citing. The figures have to be real; a page invented to look citable is worse than the original.

What about mentions on other sites?

You will read that AI tools cite brands that are mentioned often on Reddit, review sites and industry publications. It is plausible that being discussed by others helps — these systems search the web, and the web includes those places. But there is no published evidence from OpenAI or Perplexity about how mentions are weighted, and Google's AI guide calls manufactured mentions ineffective.

The safe version is the old version: earn real reviews, be listed accurately where your customers look, and contribute properly where you have something useful to say. How to get more Google reviews is a good place to start. Seeding your own brand name into forum threads is not.

How to check your citations without fooling yourself

Asking ChatGPT about your own business is the obvious first test, and it is easy to misread.

1. Use a logged-out or fresh session. Personal memory and past conversations can colour the answer.
2. Ask the questions customers ask, not "tell me about" followed by your business name. The second tests whether the system can find you by name, which is useful but not the same as being recommended.
3. Make sure web search actually ran. Look for source links. An answer with no sources came from the model's training, not from your current site.
4. Run each question more than once. Answers and sources vary between runs; one result is an anecdote.
5. Record the sources cited, not just whether you appeared. If a competitor is cited, read their page — it usually shows you what yours is missing.

How to track AI referrals

You can see some of this traffic, not all of it.

• In GA4, look at sessions by source for chatgpt.com, perplexity.ai, copilot.microsoft.com, claude.ai and gemini.google.com. Links from ChatGPT often carry a utm_source=chatgpt.com parameter, which makes them easy to spot.
• Some AI traffic arrives with no referrer at all — copied links, apps — and lands in Direct.
• Set these sources up as a custom channel group so they are not lost in Referral. How to track website conversions in GA4 covers the setup and how to tie them to enquiries.

What about llms.txt?

It comes up in every conversation about ChatGPT. Neither OpenAI nor Perplexity documents using llms.txt for search, and Google says it ignores it. llms.txt explained covers where it is genuinely useful.

Next step

Open your robots.txt now and check the three search bots in the table. If any are blocked, that is the fix with the highest return in this whole cluster. If you want the technical side and the content side reviewed together, the SEO service covers both.