How to get your business cited by ChatGPT and AI Overviews

If you want to get cited by ChatGPT, Perplexity, Gemini or Google AI Overviews, the honest answer is that there is no trick. These systems pull from pages they can reach, understand and trust, then quote the bits that answer the question best. Your job is to be that page, and to be mentioned by other sites they already trust.

I spend a lot of my week on this for Crypto Accountants, where the target is UK searches around "crypto accountant", and I built ALI, an AI assistant that answers UK crypto tax questions by retrieving and quoting HMRC guidance. Building the retrieval side taught me a lot about what makes a page easy to quote. Here is what I would do for any small business.

How do AI assistants decide what to cite?

The details differ by product and change often, but the broad pattern is the same. When an assistant searches the web for a live answer, it runs one or more queries, fetches a handful of pages, breaks them into passages, and picks the passages that answer the question most directly. Those passages get cited.

That has some practical consequences:

• If your page cannot be fetched, it cannot be cited.
• If the answer is buried in paragraph nine after a long introduction, a competitor's tidy paragraph wins.
• If your business name, location and services differ across sites, the system is less confident who you are.
• If nobody else mentions you, you are relying entirely on your own claims.

Notice that the first point is just technical SEO. Run through my WordPress technical SEO checklist first. AI visibility built on a site with indexation problems will not hold.

What does answer-first content look like?

Each page should answer one clear question, and the answer should appear in the first two or three sentences. Write it so it still makes sense if it is lifted out on its own, with no "as mentioned above" and no pronouns pointing at things the reader has not seen.

A simple structure that works:

1. A direct answer of 40 to 60 words at the top.
2. Question-style subheadings that match how people actually phrase the query ("How much does a WordPress site cost to maintain?" rather than "Maintenance").
3. A short answer under each subheading, then the detail.
4. Specifics. Named steps, named tools, conditions and exceptions. Vague advice is rarely quoted because it does not answer anything.
5. An FAQ section for the follow-up questions.

When I built ALI, every law answer had to quote the relevant section and link the exact HMRC manual page. The pages that retrieved best were the ones where a single paragraph stated a rule plainly. The same applies to your service pages. If you want an AI to say "a dental clinic in Leeds offers emergency appointments on Saturdays", there needs to be a sentence on your site that says almost exactly that.

Why does entity consistency matter?

An entity is a thing a system can recognise: your business, your founder, your product. AI tools and search engines build a picture of an entity from many sources. When those sources agree, the picture is clear. When they conflict, you get ignored or, worse, described wrongly.

Make these identical everywhere:

• Business name, written the same way each time.
• Address and phone number, if you serve a local area.
• A one-line description of what you do and who for.
• Founder or key people's names and roles.
• Links between your site and your profiles (Google Business Profile, LinkedIn, industry directories).

Audit this properly. Search your business name and look at every listing on the first two pages. Old addresses and abandoned social profiles are common, and they muddy the signal.

Which schema helps AI systems understand a site?

Structured data does not guarantee a citation, but it removes guesswork. It lets a system confirm what a page is and how facts connect, without relying solely on parsing prose.

The useful types for most businesses:

• Organization or LocalBusiness, with name, url, logo, address, telephone and sameAs pointing at your real profiles.
• Person for founders or authors, linked to the organisation.
• Service or Product for what you sell.
• Article with author and dates for blog content.
• FAQPage where the page has a visible FAQ.
• BreadcrumbList for site structure.

The rule I apply on every build is that schema must match visible content. Marking up things that are not on the page is a fast route to being distrusted by Google, and there is no reason to think AI systems will be more forgiving.

Are you blocking the AI crawlers you want?

Open your robots.txt and check. Many sites block AI crawlers by accident, through a security plugin, a CDN bot setting or a copied template. The ones most businesses care about:

• GPTBot: OpenAI's crawler. OpenAI also documents a separate user agent used when ChatGPT fetches pages for search, so check their current documentation for the full list.
• PerplexityBot: Perplexity's crawler for its index.
• Google-Extended: not a separate crawler but a control token. It governs whether Google can use your content for its Gemini models. According to Google's documentation it does not affect Google Search, and AI Overviews are part of Search, so blocking Google-Extended is not a way to opt out of AI Overviews.
• ClaudeBot and others, if you want coverage beyond the big three.

User agents and their purposes change, so check each provider's current documentation rather than trusting a list in a blog post, including this one. Then check your CDN or firewall too. A robots.txt that allows GPTBot means nothing if Cloudflare returns a challenge page to it.

Is llms.txt worth adding?

llms.txt is a proposal for a plain Markdown file at the root of your site that gives language models a curated summary and a list of your most useful pages. The idea is reasonable. The honest position is that adoption by the major AI companies is uncertain, and I have not seen clear evidence that any of them rely on it for citations.

My approach: it takes half an hour to write a good one, so I add it where it is cheap, keep it accurate, and do not count on it. If you have limited time, spend it on the answer-first rewrite and third-party mentions instead.

How do third-party mentions help?

This is the part most businesses underweight. When an assistant answers "who are the best accountants for crypto tax in the UK", it often leans on lists, directories, reviews, forum threads and news coverage, not just individual company sites. If you are absent from those, your own pages have to do all the work.

Practical places to start:

• Industry directories and professional body listings, with consistent details.
• Your Google Business Profile, kept complete and active.
• Genuine customer reviews on the platforms your sector uses.
• Guest articles, podcasts and interviews where you can explain something useful.
• Answering questions in relevant communities, honestly and without spamming links.

On the Crypto Accountants side, weekly link outreach is a fixed part of the routine for exactly this reason. You can read more about that in the Crypto Accountants web platform case study.

How do you measure progress?

Build a fixed list of 20 to 30 prompts that real customers would type. Each month, run them through ChatGPT, Perplexity, Gemini and Google, and record whether you appear, what is said and which URL is cited. Keep the wording identical so the comparison means something. Pair that with referral traffic from AI tools in your analytics. It is manual, but it is the most honest measurement available today.

Need a hand?

If you want this done properly on your own site, from crawler access and schema through to rewriting key pages answer-first, take a look at my technical SEO and AEO service or see how I approached retrieval in the ALI AI assistant case study. Then get in touch.

Hamza Naqvi

Software engineer in Lahore. 200+ websites shipped since 2020, from WordPress and Shopify builds to multi-tenant SaaS and AI assistants. More about me.