How to get found in ChatGPT

The question every business owner is starting to ask. The answer is not what most SEO agencies are selling.

Why this question matters right now

Six months ago this question was theoretical. People were still googling, and ChatGPT was a research tool and writing assistant that mostly technical users played with. Nobody treats it that way now.

In the last ninety days I have watched real people ask ChatGPT to find a contractor, book a reservation, recommend a dentist, and compare insurance quotes. They skipped Google and never visited a website. They asked an AI, read the answer, and acted on it. The business went to whoever the AI recommended, and everyone else might as well not have existed.

It is already happening, in volume, across industries where the decision can be reduced to criteria. If people currently find you by searching, reading reviews, and comparing options, an AI agent can do that same sequence in two seconds, and it will not show your customer your website unless you have built to be found by machines.

The companies that learn how to show up in ChatGPT, Gemini, Perplexity, and the next ten assistants before their competitors do will own the channel. The ones that wait may find their revenue flat while their category grows.

How ChatGPT decides what to recommend

ChatGPT does not rank businesses the way Google ranks pages, and backlink counts are not what decide who it names. Much of the playbook agencies have sold for twenty years does not carry over, and plenty of agencies are now repackaging SEO as AI visibility without changing what they do.

ChatGPT uses a training dataset, a real time search layer when it needs current information, and a reasoning model that evaluates sources the same way a person with perfect recall and unlimited time would. It weights trust signals, checks for consistency across sources, and prefers recent information to stale information. It answers based on what it can verify rather than what you claim.

When someone asks ChatGPT for a recommendation, it looks for a business it can confirm is legitimate, current, and a match for the criteria the person described. Backlink counts do not enter into it. It checks structured data, reads citations, weights reviews by recency and specificity, and confirms your facts are the same everywhere they appear. If it cannot verify you, it recommends someone it can.

So you are competing on trust more than rank. Businesses a machine can read and verify get recommended, and the rest never enter the conversation.

The three things ChatGPT looks for before it names a business

Structured facts it can parse. ChatGPT needs your name, your category, your location, your services, your hours, and your contact information in formats a machine can read. What it needs here is structured data rather than marketing copy: schema markup, business listings, citations in directories the model trusts, and content that states what you are in plain language a machine can quote without interpreting.

If your facts are inconsistent across sources, ChatGPT does not resolve the conflict. It skips you. If your site says one thing and your Google profile says another, you fail the consistency test, and the AI moves to a competitor whose information agrees.

Signals it can verify. Reviews, citations, mentions in trusted sources, and recent activity. ChatGPT weights these the same way a careful buyer would. A review from six months ago that describes a specific experience counts more than a five star rating with no text. A citation from a source the model already trusts counts more than a mention on a site it has never seen. Recent information counts more than old information, because the model assumes businesses change.

You cannot fake these signals for long. The old search engine tricks do not hold up, and the models have seen most of them. The businesses that get recommended are the ones whose actual record supports the claim.

Answers to the questions people actually ask. When someone asks ChatGPT to find a business, they are asking a question. Do you take this insurance. Are you open evenings. Can you deliver by Thursday. Do you serve this area. The businesses that answer these questions in plain text on their site, in their listings, and in their structured data get recommended. The ones that make the AI infer the answer from marketing language do not.

This is the simplest change and the one most businesses skip. Find the fifteen questions a buyer asks before they choose a company in your category. Answer them plainly, in text a machine can read and quote, rather than in a PDF or behind a contact form.

What most businesses are doing wrong

They are treating this like SEO. They are buying backlinks, optimizing title tags, and waiting for organic traffic to show up in analytics. Most of that does little for AI visibility, because the AI is not judging your site the way Google does.

They are publishing marketing content instead of structured facts. An AI cannot quote your value proposition. It can quote your hours, your pricing, and your service area if you publish them in a form it can parse. Most businesses bury that information three clicks deep or leave it out entirely because they assume a person will call to ask. An AI will not call; it recommends the competitor who listed the facts.

They are ignoring inconsistencies. Your website says you are open until 8 pm. Your Google listing says 7 pm. Your Yelp page says 6 pm. A person might call to confirm. An AI treats this as a trust signal failure and moves on. Every source a machine reads needs to say the same thing, and most businesses have never audited for this because it did not matter when people were doing the checking.

They are not building for transaction completion. Being recommended is not enough. The AI needs to finish the job. If the only way to book is a form that requires a human to read it and respond, or a phone tree that does not answer after hours, the AI cannot complete the task, so it finds a business that gave it a door and books there instead. The same pattern now applies to products, as agentic shopping in ChatGPT, Google, and Amazon shows.

The six steps to get found in ChatGPT

1. Audit your facts across every source a machine reads. Your website, your Google Business Profile, your Yelp listing, every directory, every citation. Your name, services, hours, contact information, and service area need to match exactly. One inconsistency breaks trust, and the AI skips you. Fix the conflicts before you do anything else.
2. Publish structured data on your site. Schema.org markup that tells machines what you are, where you are, what you offer, and how to contact you. AI systems deciding whether to recommend you read this markup, and if it is missing or wrong, they cannot verify you.
3. Answer the buyer questions in plain text. Find the questions people ask an assistant before they choose a company in your category. Do you take insurance. What areas do you serve. What are your rates. Can you deliver by X date. Answer every one on your site, in simple declarative sentences a machine can quote. Skip the jargon and the marketing fluff, and do not hide answers in PDFs.
4. Build recent, specific trust signals. Reviews that describe what you did and when. Citations from sources the model trusts. Content that demonstrates you are active and current. Stale information gets weighted lower than recent information, and the businesses that look dormant do not get recommended even if their facts are right.
5. Open a door the AI can complete through. This is the step that decides the sale. Can the AI book the appointment, get the quote, or place the order without a human on your side intervening? If yes, you are competing in the agentic commerce market. If no, you are competing in the call and wait market, and that market is shrinking.
6. Test whether you show up. Ask ChatGPT, Gemini, and Perplexity the questions your customers ask when they are looking for a business like yours. Use the exact phrasing a real person would use. If you are not in the answer, you are not visible, and you need to go back and fix what is missing. Run this test monthly, because the models update and your competitors are moving.

How to test whether you are visible right now

Open ChatGPT. Ask it the question your customer asks when they are looking for a business like yours. Use the real phrasing. If you are a dentist in Austin, the question is not "optimize my dental practice for AI search." The question is "find me a dentist in Austin who takes Delta Dental and is open evenings."

If your business is in the answer, you are visible. If it is not, you have work to do. Run the same test in Gemini and Perplexity. The results will differ, because the models weight sources differently, but the pattern will be consistent. The businesses that show up are the ones with structured facts that a model can verify and that are kept current. When I ran this test for 50 independent coffee roasters, ChatGPT named none of them, and the full method is in the coffee roaster study.

Do this monthly, because the models change and your competitors keep moving. Information that was current three months ago is stale now, and stale information gets weighted lower. Treat it as a channel you operate, the way you already operate search, rather than a one time project. Operating that channel every month is the job of an agentic commerce advisor.

FAQ

How does ChatGPT find businesses to recommend?

ChatGPT uses a training dataset, a real time search layer for current information, and a reasoning model that evaluates trust signals. It reads structured data, checks for consistency across sources, weights reviews and citations, and prioritizes businesses whose facts it can verify. It does not crawl websites the way Google does. It looks for information it can parse and confirm.

Does SEO help you get found in ChatGPT?

Some of it does. Structured data, consistent citations, and content that answers real buyer questions transfer directly from SEO to AI visibility. Backlinks, keyword density, and title tag optimization do not. The parts of SEO that were about being legible to machines still apply. The parts that were about gaming rankings do not.

How long does it take to appear in ChatGPT search?

If you fix the structural issues today, publish the right data, and build the trust signals, you can be visible in thirty to sixty days. The models update on their own schedules, and there is no submit button. You make the changes, you test, and you keep testing until you show up. That is faster than SEO used to be and slower than running an ad.

What is GEO optimization?

Generative Engine Optimization. The practice of making your business visible and recommendable to AI assistants like ChatGPT, Gemini, and Perplexity. It combines elements of SEO, structured data, citation building, and content strategy, but it is a separate discipline because the engines that recommend you are evaluating different signals than the engines that rank you.

Can any business get found in ChatGPT?

Any business that serves a need people describe in natural language can be found, if they do the work to be legible and verifiable. The businesses that struggle are the ones whose differentiation is subjective, whose facts are inconsistent, or whose transaction requires a human sales process. For everyone else, the path is clear. Fix the facts, answer the questions, build the trust signals, and open a door.