How to show up in ChatGPT, and why ranking first on Google won't do it for you

The short version

To show up in ChatGPT, first find out whether it knows your business accurately. Then check whether it names or cites you when people ask the questions you want to be found for. A Google ranking alone can't tell you either of those things.

There are two routes to an answer. ChatGPT may use information it learned during training, or it may search the web and use current sources. Being known and being cited are different outcomes. If it gets your business wrong, look for missing or conflicting facts online. If it knows you but leaves you out of relevant answers, look at which businesses and sources it chooses instead.

Why Google rankings don't carry over

Google's first page is a list of results for a search query. ChatGPT produces an answer to a question, sometimes using a live search and sometimes relying on what the model has learned. I see ChatGPT-style models as drawing on a far wider pool of human knowledge than Google's index alone, so I wouldn't expect them to choose sources by copying Google's first ten results.

I see this in work with a digital agency for which I provide white-label support. The agency has good organic visibility for several target sector marketing terms. Yet when those terms become questions in ChatGPT, the agency is cited much less often. ChatGPT sometimes references agencies that don't appear on Google's first page for the related keywords. Those citations also change more from one conversation to the next than a typical set of Google results.

That observation doesn't reveal a fixed ChatGPT ranking formula. It does show why checking your Google positions is a poor substitute for checking ChatGPT's answers. BrightEdge's study of five AI engines found that pairs of engines shared between 16% and 59% of their top 100 cited sources. Those figures compare AI engines with each other, not ChatGPT with Google's organic results, but they show how much source selection can differ. I unpack the study in the 14% problem.

Route one: what ChatGPT already knows

A model can answer from patterns and facts it learned during training. OpenAI distinguishes those answers from answers that use web search. Training isn't a live check of your website, so a correct page published today won't automatically correct every answer based on earlier information.

Start with direct questions. Ask ChatGPT what your business does, where it operates and which services it offers. Write down the claims it makes, including omissions and mistakes. Repeat the check in fresh conversations so one unusually good or bad answer doesn't set your priorities.

If a fact is wrong, check the information available for ChatGPT to encounter. Is your own site clear? Do relevant independent sources describe the same business, services and location? Are old descriptions still prominent? You can't edit a trained model's memory on demand, but you can make the public record clearer and more consistent for future training and retrieval.

Correct recall is useful, though it isn't the same as visibility for a generic question. ChatGPT might describe your agency accurately when asked by name and still leave it out of an answer about agencies in your sector. That answer has limited room, and other businesses are competing for it.

Route two: what ChatGPT fetches

ChatGPT can also search the web for current information. When it does, an answer may include links to the sources it used. A citation tells you something about that answer's source selection. It doesn't prove that the model had never heard of your brand before, and a mention without a citation doesn't prove that it fetched your site.

For this route, look beyond your own pages. For the questions that matter to your buyers, note which sources ChatGPT cites and whether they contain accurate facts about you. In my experience, a wider web footprint matters: a business described and sourced elsewhere has more opportunities to appear in the material behind an answer. The exact sources can change between conversations, so treat one citation as a clue rather than a permanent position.

Schema markup is important because it helps machines understand what your pages describe. Clear site content matters too, but schema on its own isn't a shortcut to a mention or citation in ChatGPT or other LLMs. ChatGPT has already learned from ordinary web content, and a live search can find ordinary pages too. The harder question is whether enough trustworthy, relevant material about your business exists, including beyond your own site. My GEO service starts with that question.

What to do first, second, third

1. Ask and record. Ask ChatGPT about your business by name. Then ask several questions a prospective customer would ask without naming you. Save the wording, date, answer, mentions and cited sources. Repeat the questions in fresh conversations.

2. Fix the facts. For each wrong or missing fact, check your website and relevant independent sources. Make your name, services, location and other important details clear and consistent. Prioritise information a customer needs to make a decision.

3. Build a useful footprint. Look at the sources ChatGPT chooses for your sector. Where relevant, seek coverage, references or inclusion that gives readers a reason to learn about your business. Recheck the same questions after you make changes, while allowing for variation between answers.

I'd save the money offered for inserts in paid “best of” lists. I'd also leave Reddit manipulation alone. And I wouldn't expect an llms.txt file to unlock ChatGPT visibility. I've written about what that file can and cannot do. None of those shortcuts replaces accurate facts and a credible presence elsewhere online.

Can you use ChatGPT for your own SEO?

ChatGPT can pull together plenty of useful recommendations to work from. SEO isn't an exact science, though, so there's plenty of conflicting advice. Some people seek less reputable routes to success, often called black hat SEO. White hat techniques are the safer bet, even if they take longer to show results.

ChatGPT can present advice confidently without a definitive understanding of your site's strengths and weaknesses, or which changes would follow best practice for that site. It also doesn't have access to the data SEO professionals pay a lot of money for, such as search volume data, keyword trends, site crawlability and technical adherence.

How to check where you stand

Keep two columns in your notes: what ChatGPT says about you and when it names or cites you for relevant questions. This separates an accuracy problem from a visibility problem. Include the sources behind cited answers, because they may point to a gap you can address.

Run each question more than once. Record the exact prompt and use fresh conversations. A single answer is a sample, not a ranking. If you change your site or earn new coverage, compare later samples with the earlier ones and be honest about what the small set can show.

You can start with the free AI visibility audit. It gives you a way to see what assistants say about your brand and where it appears. From there, the useful question is specific: which facts need correcting, and which customer questions are you absent from? That gives you work to do, rather than another number to watch.

I expect models to get quicker at discounting tactics designed to game their answers. The fall in Reddit citations points in that direction, though it doesn't prove why they fell. If marketers exploit another site that starts getting cited more often, I expect its value as a shortcut to fall too.

ChatGPT is worth checking early because of its reach, but reach isn't a citation count. The difference between chatbot use and traffic sent to websites is covered in usage is not citation. For your own business, the place to begin is simpler: ask the questions, keep the answers and follow the evidence.