Why your business isn't showing up in ChatGPT, and what to fix first

How to check today, before you fix anything

Start with what customers would see. This takes ten minutes and costs nothing.

Open ChatGPT in a fresh chat. Ask the questions your customers ask, worded the way they would say them, with your city in it. A plumber in Clearwater might try “who should I call for a burst pipe in Clearwater” and “how much does a water heater replacement cost in Pinellas County.” Write down every business it names and the reason it gives for each. Ask each question a second and third time, in different words. Answers change from one ask to the next, so the names that keep coming back tell you more than any single reply.

Then do the same for your own name. Ask what ChatGPT knows about your business. If it gets your services, city or phone wrong, that is a clue. It is reading something outdated or contradictory, or it has nothing to read at all.

Next, run the free AI visibility check. It fetches your site the way an AI crawler does and reports what came back: whether robots.txt lets the main AI crawlers in, OpenAI's GPTBot and OAI-SearchBot among them, whether your page text exists in the HTML without JavaScript, whether the page answers questions, and whether structured data, a sitemap and an llms.txt file are present. It takes seconds and needs no email. It does not check reviews or listings, and it cannot tell you whether the engines name you today, because none of that can be read from one page. That is what the questions you just asked are for.

Now you have two pictures: who ChatGPT names, and what your site looks like to a machine. The rest of this guide explains the gap, in the order to close it.

Reason 1: AI can't read your site

This comes first because it can cancel everything else. A site that is blocked, empty to a crawler or marked “do not index” gets no benefit from good copy or good reviews. There are three common causes.

Your robots.txt blocks OpenAI's crawlers

Go to yourwebsite.com/robots.txt. It is a plain text file, and anyone can read it. Look for lines that name a crawler under “User-agent” followed by “Disallow: /”. Also look at the general rule for all crawlers, written as “User-agent: *”. Website builders, security plugins and well-meaning developers sometimes add these blocks without telling the owner.

OpenAI documents its crawlers on its bots page. Three matter to a local business:

• OAI-SearchBot is for search. OpenAI says sites that opt out of it will not be shown in ChatGPT search answers, though they can still appear as navigational links. This is the one to allow.
• GPTBot collects content that may be used to train OpenAI's models. OpenAI says disallowing it signals that your content should not be used for training. Blocking it is a choice about training, not about search. Some owners block it on purpose, and that is a legitimate call.
• ChatGPT-User visits a page when a person using ChatGPT asks for it. OpenAI says it is not used to crawl the web automatically or to decide whether content appears in search, and that robots.txt rules may not apply to it.

One detail about the free check: it reports a block on any of seven AI crawlers, GPTBot and OAI-SearchBot among them. If it flags GPTBot and you blocked it on purpose, that is your call to make. A block on OAI-SearchBot is the one to fix.

If you change the file, be patient. OpenAI says it can take about 24 hours from a robots.txt update for its systems to adjust. OpenAI also recommends allowing requests from its published IP ranges, which matters if a firewall or security service sits in front of your site. Ask whoever runs your hosting to check that.

Your text only appears after JavaScript runs

Some websites send the browser an almost empty page and build the content afterward with JavaScript. You see a finished page. A crawler that does not run that code may see a blank one. You can test it in a minute: open your page, right-click, choose “View Page Source” (not “Inspect”), and search for your phone number or a sentence from your services section. If it isn't in the source, a crawler that doesn't run JavaScript won't find it either. Heavy sliders, tabs and pop-up menus that load their text on click are common culprits. The fix is to put the words in the page itself.

A noindex tag tells search to skip the page

A noindex tag is a line in a page's HTML that asks search engines not to list it. It often gets left behind from the days a site was being built. Search the same page source for the word “noindex.” If you use WordPress, check Settings, then Reading, for the box that discourages search engines from indexing the site. If either is on, turn it off and ask your web person why it was there.

Reason 2: your site doesn't answer the question in plain text

Once a crawler can get in, it needs something worth quoting. Many local business sites say who they are and how to reach them, then stop. A customer asking ChatGPT about your trade is asking something specific, and a page that answers it directly is easier to quote than one that doesn't.

Open your site as a stranger and look for plain-text answers to the things people ask before they call:

• What it costs. You don't need a price list if your pricing is quoted per job. Explain how you price: what changes the number, what a typical visit includes, and how a quote works. “Call for pricing” alone is an answer nobody can quote.
• Where you work. Name the cities and neighborhoods you serve, in a sentence, not only in a map graphic or a footer list.
• Insurance, licensing and payment. Which insurance plans you accept, if it applies. Which licenses you hold. How people pay.
• Hours and availability. When you answer the phone, whether you do emergencies, and how fast you usually reply.
• What is included. What a customer gets for a visit, a treatment or a package, and what costs extra.

Write each answer as a short paragraph under a heading that sounds like the question, with the answer in the first sentence. An answer buried in the fourth paragraph or hidden in an image is hard to quote.

One accuracy note. Don't write a number you can't stand behind. If your price depends on the job, say what it depends on. A true “it depends on these three things” beats an invented range.

Reason 3: nothing else backs you up

The third test is whether other sources agree with what your website says. A site that describes itself and has no outside confirmation is a claim. A site that matches your Google Business Profile, your reviews and the directories your trade uses is easier for anything, human or machine, to trust. No one outside OpenAI knows exactly how ChatGPT chooses names, and OpenAI doesn't publish a formula. This one rests on a simple principle: more independent sources telling the same story is a stronger signal than one.

Here is where to look:

• Google Business Profile. Claim it, fill in every field, pick the right categories, and keep your hours current.
• Reviews. Ask every happy customer, right after the job, while it is fresh. Never buy or write reviews yourself. Google prohibits fake reviews, and they put your whole profile at risk.
• Listings that agree. Your business name, address and phone number should read the same on your site, your Google profile, Facebook, Yelp and the directories for your trade. If you serve customers at their homes and don't have a storefront, use the same service area everywhere instead. Small differences, like “Street” in one place and “St.” in another, matter less than a wrong phone number or an old address.

What not to waste money on

Plenty of money gets spent in this area, and some of it is wasted. Here is what to be careful about.

• Anyone who guarantees a position. No one controls ChatGPT's answers, and they change. A promise to put you first in ChatGPT is a promise nobody can keep. Ask what happens if it doesn't work.
• Reports that tell you the problem and not the fix. A score is only useful if it points at something you can change. Monitoring tools that track whether you get named can be worthwhile once the basics are in place. Before that, they mostly count a gap you already know exists.
• Bulk content before the site is readable. Fifty new articles on a site that blocks crawlers, or that hides its text behind JavaScript, add nothing. Fix the access problem first.
• Mass directory submissions. A handful of correct, matching listings does more than hundreds of low-quality ones with wrong details. Accuracy is the point.
• Fake or bought reviews. Covered above. They also break the whole idea of the exercise, which is that other people confirm what you say.
• An llms.txt file as the whole plan. It is a cheap, useful file that tells AI tools who you are, and we put one on every site we build. It does not fix a blocked robots.txt or a page with no readable text. Treat it as an addition, not a rescue.

A fix-first checklist

Work down the list in order and stop to retest when you finish each step.

• Ask ChatGPT your customers' questions for your city. Write down who it names.
• Run the free check on your homepage and on your main service page.
• Open yourwebsite.com/robots.txt. Make sure OAI-SearchBot is not blocked. Decide on GPTBot deliberately.
• View the page source. Confirm your services, city and phone number are in the HTML. Search for “noindex.”
• Add plain-text answers for cost, service area, insurance, hours and what is included.
• Fill out your Google Business Profile. Make your name, address and phone number match everywhere.
• Start asking for a review after every job.
• Wait, then ask ChatGPT the same questions again. After a robots.txt change, allow at least a day.

That is the same order we use in our AI visibility work: make the site readable, write the answers, build the proof, then measure. If you want help with any of it, you can book 15 minutes with Mike. He will look at your actual site, not a template.