How to get your business to show up in ChatGPT
There is no form to submit your business to ChatGPT, and no setting to switch on. Here is what actually decides whether it names you.

Somebody asked ChatGPT for the best plumber in your town and it named four businesses. None of them was yours. The obvious next question is where you go to fix that, and the honest answer is the one nobody selling AI visibility services wants to lead with: there is nowhere to go. There is no form, no dashboard, no submission process and no paid placement.
Our articles are still written by humans!
Get human written articles in your Google feed.
| How the answer was produced | What decides whether you appear | Can you influence it | How quickly |
|---|---|---|---|
| From training | How often and how consistently your business was described across the web before the model was trained | Only slowly, and only by being written about | Months at best, and only at the next training cut-off |
| From a live search | Whether your pages are crawlable, indexed and clearly answer the question asked | Yes, directly | Days to weeks |
| From a link the user pasted | What is in the page text at the moment it is fetched | Yes, completely | Immediately |
There is no form, and there is no listing
This is worth saying flatly because a small industry has grown up around implying otherwise. OpenAI does not operate a business directory. There is no equivalent of a Google Business Profile inside ChatGPT, no verification step, no category to choose and nothing to claim. When ChatGPT names a business it is either repeating something it absorbed while being trained, or it has just run a search and is summarising the pages it found.
If somebody offers to get your company listed in ChatGPT for a monthly fee, ask them which of those two things they are changing. In practice the answer is usually that they will add you to a set of directories, which is a real if modest tactic, described honestly further down this page, and not a listing in ChatGPT.

The two ways ChatGPT ends up naming a business
From what it learned
Ask a general question with no browsing and the model answers from what it absorbed during training. For a business to be in there, it has to have been described enough times, consistently enough, across enough of the public web, that the association survived. For a national brand that is routine. For a five person firm in one town it is unusual, and it is not something you can arrange in a quarter.
OpenAI's documentation names the crawler behind this half as GPTBot, described as the one that will crawl content that may be used in training generative AI foundation models. Note the tense. Anything it collects affects a future model, not the one answering today.
From a search it ran while answering
Most questions that name a place or ask for a recommendation now trigger a search, and the answer is assembled from pages fetched at that moment, usually with citations. This is the half worth working on, because it behaves much more like search always has. Pages that are crawlable, clearly written and actually answer the question get quoted. Pages that do not, do not.
The crawler for this half is a different one, and confusing the two is the single most expensive mistake we see.
Why your business is not showing up
In our experience the reason is almost always dull, and almost always one of these six.
- Your robots.txt blocks the crawler that feeds ChatGPT's search results, usually because it was added during the wave of AI blocking a couple of years ago and nobody revisited it.
- The facts an assistant needs are on your site only as pictures. Opening hours inside a poster graphic, prices inside a PDF menu, the service list inside a hero image with text baked into it.
- Nothing outside your own domain describes what you do. A business that appears nowhere except on its own website is a business with one source, and one source is thin evidence.
- You are described differently in different places. Three variations of the company name, two addresses, an old phone number on a directory nobody has looked at since 2019.
- Your pages talk about your values and your journey rather than what you sell, to whom, where, and for how much.
- You are genuinely new, or genuinely small, in a category with a hundred alternatives. This is a real answer and it deserves saying rather than being papered over with tactics.
Check what it says about you today
Do this before changing anything, and write the answers down. It takes about ten minutes and it is the only baseline you will get. If you would rather have something do it for you, we have compared the tools that automate exactly this, with what each costs and where the free option runs out. What those tools cost, tier by tier, is set out separately: the published range runs from 29 euros a month to 800, and the cheapest plan covers more AI engines than one well known product's 99 dollar one.
| Ask this | What you are looking for |
|---|---|
| What do you know about [your business name] in [town] | Whether it exists in the model at all, and whether the description is right |
| Who are the best [your category] in [town] | Whether you appear unprompted, and who does instead |
| Compare [your business] and [your closest competitor] | Which facts it reaches for, and which of them are wrong |
| What does [your business] charge for [your main service] | Whether pricing information exists anywhere it can reach |
| Cite your sources for that | The pages it is actually reading, which is the most useful output of the whole exercise |
Let the right crawler in
This is the one genuinely technical item on the list, and it is worth doing first because it is binary. OpenAI publishes four separate agents and they do different jobs. Blocking the wrong one removes you from ChatGPT's search results entirely, which is a much bigger penalty than most site owners realise they opted into.
| Agent | What it is for | Should a small business allow it |
|---|---|---|
| OAI-SearchBot | Surfacing websites in search results in ChatGPT's search features | Yes. Blocking this is what keeps you out of answers |
| GPTBot | Crawling content that may be used to train future models | Usually yes, if you want to exist in future models. This is a genuine judgement call about your content |
| ChatGPT-User | Fetching a page because a user asked for it. OpenAI states it is not used for crawling the web in an automatic fashion | Yes. Blocking it stops people pasting your link and getting an answer |
| OAI-AdsBot | Validating the safety of pages submitted as ads | Only relevant if you advertise there |
Open your own robots.txt in a browser and read it. If it disallows any of the first three, decide deliberately whether you meant to. The distinction that catches people out is that allowing GPTBot does nothing for how you appear today, and blocking OAI-SearchBot damages it immediately. Several sites did exactly the wrong pair.

Be described the same way everywhere
An assistant assembling an answer about a small business is reconciling several sources. Where they agree it states the fact confidently. Where they disagree it either hedges or picks whichever source it trusts more, and that will not always be you.
So the unglamorous work is real work: one spelling of the business name, one address, one phone number, one description of what you do, repeated identically on your website, your Google Business Profile, your social profiles and the three or four directories that actually cover your industry. Not similar. Identical.
Get named on pages you do not own
This is the part with the highest ceiling and it is not a trick. Assistants weight independent description heavily, because a business describing itself is not evidence of anything. A local news mention, a supplier's stockist page, a trade association member list, a genuine review on a site your industry actually uses, a partner's case study. Each one is a source that agrees you exist and says what you do.
One caveat worth stating, since this is where the money gets wasted. Bulk directory submission packages produce dozens of near identical pages on sites nobody reads, and assistants discount them for the same reason people do. Four relevant mentions beat sixty irrelevant ones, and the sixty carry a real risk of introducing inconsistent details that then need cleaning up.
Answer the question in the words people ask it
Pages that get quoted tend to state an answer plainly, near the top, in the language of the question. Pages that get skipped bury the answer under positioning. If people ask how much something costs, the page should contain a number or a range and the reason for it, in text.
This is not a new discipline. It is the same instinct as writing something genuinely useful, which Google has been describing for years in its guidance on creating helpful, people-first content. The change is that being the clearest answer now gets you quoted as well as ranked.

Put the facts in text
Prices in a PDF, hours in a poster image, the service list rendered as part of a hero graphic. All of it is invisible to the thing reading your site. Google's own advice on AI features is to make important information available as text, supplemented by images rather than replaced by them, and the same applies to every assistant.
What does not work
Being specific about this is more useful than another list of tactics, because most of the wasted effort we see is enthusiastic rather than lazy.
- Special files for AI crawlers. Google states directly that site owners do not need to create new machine readable files, AI text files or markup to appear in its AI features. Adding one is cheap and harmless, but treat anyone presenting it as the answer with caution.
- Writing best plumber in Manchester forty times across your site. Repetition was a weak signal for search engines and it is a weaker one for a model reading for meaning.
- Paying for AI visibility listing packages. Ask precisely what changes. If the answer is directory submissions, price it as directory submissions.
- Generating a hundred thin pages to cover every phrasing. Volume without substance is the exact pattern both Google's helpful content guidance and every assistant's summarisation step are built to discount.
- Publishing on your own site that you are the best. No assistant treats self description as evidence, and neither does anyone else.
Where social media actually fits
The useful part is narrow. Your profiles are public pages that state your name, your location and what you do, and they are usually crawlable, so they contribute to the consistency described above. Fill them in completely and make sure they say the same thing as your website. An abandoned profile with an old address is a source of contradiction, which is worse than no profile at all.
What social media does not do is substitute for the rest of this list. If you want the honest version of what it does and does not achieve for a small business, we wrote it in is AI social media worth it for a small business, and the more general question of what to publish is in what a small business should actually post about.
How long any of this takes
For the search half, allow weeks rather than days, and expect it to move roughly in step with your ordinary search visibility, because it depends on the same crawling and indexing. Fixing a robots.txt block is the exception, since that can change what is available almost immediately.
For the training half, nobody outside OpenAI can tell you, and the honest expectation is that a small local business may never be in there in a meaningful way. That is not a failure. Being reliably found by the search half is the achievable goal, and it is where the enquiries come from anyway.
The short version
- Read your robots.txt today and make sure OAI-SearchBot and ChatGPT-User are not blocked.
- Run the five questions in the table above and save the answers, including the cited sources.
- Fix every inconsistent name, address, phone number and description across your own site, your Google Business Profile and the directories your industry actually uses.
- Move prices, hours and service lists out of images and into text.
- Earn three or four genuine mentions on sites you do not own, and stop there rather than buying sixty.
- Rewrite your two most important pages so the answer comes first and the positioning comes second.
Frequently asked questions
How do you rank in ChatGPT?
There is no ranking to win, which is the first thing to understand. ChatGPT does not order results, it writes one answer and either names you in it or does not. What moves that is being crawlable by OAI-SearchBot, answering real questions in plain text on your own pages, and being described the same way by other sites, so the model keeps meeting one consistent account of what you do.
How do I get my business listed in ChatGPT?
You cannot, because no listing exists. ChatGPT has no business directory, no submission form and no verification process. It names a business either from what it absorbed during training or from a search it runs while answering. The second of those is the one you can influence, by making sure your pages are crawlable by OAI-SearchBot, that they answer questions in plain text, and that other sites describe your business the same way you do.
Why is my business not showing up on ChatGPT?
Usually one of six things: your robots.txt blocks the crawler that feeds ChatGPT's search results, your key facts exist only inside images or PDFs, nothing outside your own website describes you, you are described inconsistently across the web, your pages talk about values rather than what you sell, or you are genuinely new in a crowded category. Check the robots.txt first, because it is the only one that is a straight switch.
Should I block GPTBot in robots.txt?
That is a real judgement call, and it is separate from appearing in answers. GPTBot collects content that may be used to train future models, so blocking it affects whether you exist in a model released later, not how you appear today. OAI-SearchBot is the one that surfaces pages in ChatGPT's search features, and blocking that removes you from answers now. Blocking OAI-SearchBot while allowing GPTBot is the worst of both.
Do I need an llms.txt file to appear in AI search?
No. Google states directly in its documentation on AI features that site owners do not need to create new machine readable files, AI text files or markup, and that there are no additional requirements or special optimisations to appear in AI Overviews or AI Mode. Adding one costs almost nothing, but it is not the thing standing between you and being mentioned.
Is generative engine optimisation different from SEO?
Less than the name suggests. The crawling, the indexing and the judgement about whether a page is worth quoting are largely the things search already rewarded. What changes is the emphasis: plain answers near the top of a page, facts in text rather than in images, and consistent independent descriptions of your business matter more when an answer is being summarised than when ten links are being ranked.
Does posting on social media help my business appear in ChatGPT?
Only in a limited way, and we sell a social media tool so weigh that accordingly. Complete, current, crawlable profiles are one more source that agrees on your name, location and what you do, which helps consistency. Posting frequency itself is not a meaningful factor, and an abandoned profile carrying an old address actively hurts by contradicting your other sources.
Can I pay to appear in ChatGPT answers?
Not for organic recommendations. OpenAI operates an advertising validation crawler, so paid placement exists as a product area, but it is separate from being named when someone asks for the best business in a category. Anyone selling monthly AI listing packages should be asked exactly which crawlable page changes as a result, and priced on that answer.
How long before changes show up in ChatGPT?
For anything driven by live search, weeks rather than days, on roughly the same timescale as your ordinary search visibility, with the exception of unblocking a crawler which can take effect much faster. For what the model itself knows without searching, there is no published timescale, and a small local business should not plan around getting there.
How we researched this
- The technical claims in this piece come from the two companies concerned rather than from the SEO commentary around them. The crawler names, what each one is for, and the fact that one of them is not a crawler at all are taken from OpenAI's own bots documentation, opened by hand on 9 August 2026. Google's position on what a site needs in order to appear in AI features is quoted directly from Google Search Central on the same day.
- That sourcing decision is the point of the article. Almost everything written about this subject is inference dressed as method, and the two organisations that actually decide the outcome have both published plainly worded documentation that most of it contradicts.
- The checking procedure in the middle of the piece is one we run for our own customers before and after we change anything. It is deliberately manual and deliberately cheap, because paid AI visibility dashboards mostly sell a prettier version of the same twenty prompts typed by hand.
- We have not measured how long any of this takes to work, and we say so in the article rather than printing a number we cannot defend. Nobody outside OpenAI can observe a training cut-off, and any article giving you a confident timeline for that half of the answer is guessing.
- We sell a tool that publishes social content, which is one of the inputs discussed here. That interest is disclosed in the article at the point it becomes relevant, and the section about social media is deliberately the least enthusiastic one in the piece.



