How to get cited by ChatGPT and Google AI Overviews: the entity and content playbook

For most of the last twenty years, the goal of search was a click. You ranked, someone clicked, they landed on you. The answer engines have changed the shape of that. When a customer asks ChatGPT how much a service costs, or types a question into Google and reads the AI Overview at the top, a machine reads across many sources and writes a single answer, naming only a few of them. Being one of those named sources is the new version of ranking on page one, and you cannot rank for a sentence a model cannot extract or trust.

This is the practical playbook for getting a UK small business cited inside ChatGPT, Google AI Overviews and the other answer engines, sometimes called generative engine optimisation, GEO, or AI SEO. It sits under our wider guide to AI search optimisation, which covers the whole landscape. If you want the foundations and the honest caveats first, start there. If you came here to know exactly what to do, read on. There are four jobs, in order, and a fifth that tells you whether the first four are working.

Why this matters now

The reason any of this is worth your time is a shift in how people find answers, and the numbers are not small. Roughly 60 percent of Google searches now end without a click on any result, because the searcher gets what they came for on the results page itself (SparkToro). At the same time, ChatGPT reports around 900 million weekly active users (OpenAI), a large share of whom now ask it the kind of question they would once have typed into Google.

In both cases the same thing happens: a machine writes the answer and names a handful of sources. If your business is one of them, you are visible at the exact moment of the question. If it is not, you are invisible for that search no matter how well your page would have ranked in the old list of blue links. That is the whole game, and the four jobs below are how you play it honestly.

A blunt note on what you can and cannot control

Before the playbook, the honest part, because credibility is the whole point of a piece like this. Nobody outside OpenAI, Google and the other providers can promise you a citation. The mechanics are undisclosed, they change, and anyone who guarantees you a spot in ChatGPT is either mistaken or selling something.

What you can do is shift the odds, and you do it with the same fundamentals that make a genuinely good website. There is no separate “AI SEO” product to buy that bolts citations onto a weak site. You do not control the citation. You control the four things that make the citation likely, and that is the only honest promise anyone can make. Here they are.

Job 1: Write answer-first, extractable content

The first job is to make your facts trivially easy for a machine to lift. Answer engines reward clarity and skip waffle, so a page that hides its point behind four paragraphs of marketing copy will be passed over for a clearer rival.

In practice that means three habits:

• Put the answer near the top. State the key fact in the first sentence under each heading, before any elaboration. If a person skimming cannot find the answer in five seconds, a machine summarising the page will struggle too.
• Use question-shaped headings. Structure pages around the real questions customers ask, such as “How much does a small-business website cost in the UK?”, and answer each one directly underneath. A genuine FAQ section, each question answered in one or two tidy sentences, is one of the most reliable ways to feed an answer engine exactly what it needs.
• Write short, quotable sentences with clear claims and figures. A clean, self-contained statement that pairs an assertion with a specific number, and a source where you have one, is far easier to lift and attribute than a long sentence tangled with conditions. “Our emergency call-out covers Leeds and LS postcodes, with a typical two-hour response” beats a paragraph that buries the same fact.

A useful discipline: read each heading, then read the sentence directly under it. If that one sentence does not answer the heading on its own, rewrite it until it does. This is the same instinct that wins featured snippets, and it is the exact shape an AI answer borrows. The four ways small-business pages fail this test are set out in the companion piece on why small businesses are invisible in Google’s AI Overviews.

Job 2: Build your brand entity

Answer engines are more likely to cite a brand they recognise and a fact they can verify against other sources. The technical name for that recognisable, stable identity is a brand entity, and you build one by being consistent and verifiable everywhere your business appears.

Three things do most of the work:

• Consistent name, address and phone (NAP). Write your business name, address and phone number in exactly the same format on your website, your Google Business Profile and every directory, rendered as text rather than baked into an image. Consistency helps a model build a single, clear picture of who you are and trust the facts attached to it. Conflicting details make you look like two half-known businesses instead of one trusted one.
• Schema markup that labels your facts. Structured data is code in the background of your page that explicitly tells machines “this is the business, this is the address, this is the price, this is the answer to this question.” Organization or LocalBusiness site-wide, FAQPage on genuine question-and-answer content, and Article with a named author on guides. Google is clear that this does not directly rank you, but it removes the guesswork about what your page asserts. The detail is in the companion guide on schema for AI answer engines.
• A real About and credentials page. Name the people, their experience and their proof: qualifications, years in trade, genuine results. First-hand experience is exactly what a generic competitor scraping the web cannot fake, and it is what turns an anonymous page into an attributable source a model can safely name. Our own credentials page is the working example of what that looks like in practice.

This is not vanity. It is the difference between a page a model reads as a trustworthy entity and one it reads as anonymous filler.

Job 3: Earn presence in the sources models read

Here is the part most people miss: your own website is not the only thing the models read. Answer engines draw heavily on third-party sources, so getting named in the places they trust gets you into the answer by proxy.

Three honest routes, in roughly the order of effort to reward:

• Reputable roundups and listicles. Answer engines frequently lift their recommendations from “best of” articles, so being featured in a respected “best plumbers in Leeds” style roundup can get you named inside the AI answer without the model ever quoting your own site. Earn these the legitimate way, by being genuinely good and worth listing, not by buying placements.
• Relevant directories. Being present, with consistent NAP, in the directories that matter for your trade and area reinforces your entity and gives the models more corroborating sources for your facts. Quality over quantity: a handful of respected, relevant directories beats a hundred spammy ones.
• Forums like Reddit, done honestly. Large models are trained on and actively cite discussion from communities like Reddit, so authentic, helpful contributions where your expertise genuinely shows can put your name into the sources the models draw from. The word that matters is honestly. Drive-by self-promotion is obvious, against the rules of every community, and does more harm than good. Be useful first; the visibility is a by-product, never the opening move.

None of these are quick wins, and that is the point. They build the corroboration that makes a model confident enough to name you, and corroboration cannot be faked overnight.

Job 4: Measure your share of voice in AI answers

The fifth thing tells you whether the first four are working. Share of voice in AI answers means how often, and how accurately, your business is named across the main answer engines for the questions that matter to you. You cannot improve what you do not measure.

The method is unglamorous and it works: write down your ten or twenty most important customer questions, then ask them, on a regular schedule, across ChatGPT, Google’s AI Overviews, Gemini and Perplexity. Record three things each time, whether you are named at all, whether the facts about you are correct, and which rivals are named in your place. Repeat monthly. Over a few cycles you will see where you are being cited, where your facts are being quoted wrongly (which is its own problem to fix), and where a competitor has taken a slot you should own.

This is a manual check, and being honest, it is exactly the kind of recurring, easy-to-forget job that quietly slips. It is also the only way to know whether your effort is landing rather than guessing.

How this fits together as a build

Put plainly, there is no AI-citation trick. There is a well-built, clearly-written, consistently-described site that earns corroboration elsewhere, or there is not. The four jobs above are the same fundamentals that make a page good for a human searcher, which is precisely why they make it good for an answer engine too. The full nine-discipline picture, including AI Search SEO alongside the rest, is in the free SEO playbook.

For most UK small businesses the honest blocker is not knowledge but time: the content has to be answer-first, the schema has to be clean and current, the entity has to stay consistent across every listing, and the share-of-voice check has to actually happen each month. That is the work, and it does not stop. It is exactly what a managed website service is for: one person who builds the site properly and keeps these foundations current, so you are not tracking every shift in a fast-moving field yourself.

Where to start

If you want to know how your current site stands on the things that genuinely decide whether a model cites you, structure, speed, clarity and credibility, the fastest first step is a free Site Score. It tells you in plain English where you are strong and where you are leaking, before any commitment. Or if you would rather talk it through, get in touch and we will look at your site together, tell you which of the four jobs you are losing on, and which of this is worth your time and which is noise. There is no lock-in and nothing to commit to until you are sure.