How to Rank in ChatGPT for Local Businesses

Getting a local business recommended in ChatGPT comes down to making your services, location and credibility easy to find and easy to verify. That means a useful website, accurate listings, real evidence of your work, other people saying legitimate things about you, and answers to the questions customers actually ask.

Then you measure whether any of it brings in real inquiries.

That sounds suspiciously like good marketing. Sorry. I was hoping to sell you a ceremonial AI ranking crystal, but apparently we still have to do the work.

The opportunity is real, especially when your competitors barely explain what they do. But “rank in ChatGPT” needs some unpacking before somebody promises you position one by next Tuesday.

What “ranking in ChatGPT” actually means

When someone asks ChatGPT to find a local business, a few different things can happen. It can mention your company, include you on a shortlist, or cite a page from your website. Those are different outcomes. ChatGPT citing your helpful guide isn’t the same as recommending that someone hire you.

I’m focused here on ChatGPT’s web search answers, since that’s where current local information comes from. OpenAI says ChatGPT can search the web, link to sources, and use location information to find local results. Your approximate location, saved memories and the exact wording of the question all change the answer.

So “bathroom remodelers in Fort Worth who work around families still living in the house” is a completely different test from “best contractor,” especially in a chat that already knows your business name. There’s no universal leaderboard you can screenshot and retire on.

(And no, this page not showing up when you search “how to rank in ChatGPT” doesn’t disprove anything. I’m not trying to win that keyword. I’m trying to help local businesses win theirs.)

One ground rule before we start

Almost everything public about local rankings comes from Google, which publishes guidance about relevance, distance, prominence, reviews and structured data. OpenAI publishes much less about how ChatGPT picks local businesses. So when I mention Google’s guidance below, I’m using it as reasonable evidence about what makes a business easy to evaluate. I’m not claiming it’s ChatGPT’s formula. Nobody outside OpenAI knows that formula.

Start with the questions that lead to business

Pick one service you want more work for and the areas you genuinely serve. Then write down what a customer might ask before choosing someone.

For a remodeling contractor:

• Who does small bathroom remodels in Fort Worth?
• Which contractors explain how they handle dust while the family stays home?
• What should I ask before hiring someone to replace a shower?

For a massage business:

• Where can I book an evening massage near the Cultural District?
• Which local studios have quiet rooms and explain their appointment options?

These are research prompts, not search volume data. The best source is your actual phone calls, emails and sales conversations. (Just don’t paste private customer details into an AI tool without permission.)

Keep it manageable. You don’t need 600 prompts and a dashboard that requires its own IT department. Five to ten good scenarios will tell you plenty.

For each one, ask: what would someone need to know to choose us confidently, and where could they double-check it?

Make your website answer the obvious questions

A service page should make it clear what you offer, where, who it’s for, and how someone takes the next step.

“Quality solutions with exceptional service” tells me approximately nothing. A toaster could put that on its LinkedIn profile.

A bathroom remodeling page could instead cover the project types you take on, how consultations work, how you handle occupied homes, and what drives the price. Only promise things your team will actually do.

Every key page should have:

• Your real business name, phone number and service area
• Current hours and the best way to get a quote or book
• Qualifications, with a way to verify them (like a license number people can look up)
• Real project examples, your own photos, and specific explanations of the work
• Clear boundaries, like services you don’t offer or areas you don’t cover

Put the answers in actual text. Photos prove you do good work, but a photo can’t explain your process or tell someone whether you serve their zip code.

Location pages are fine when they contain genuinely local information. Swapping the city name on 80 otherwise identical pages isn’t. Google specifically warns against doorway pages like that, and they don’t help a customer decide anything.

Make sure search crawlers can actually get in

Before you publish anything new, have your developer check your important pages for accidental crawl blocks, noindex tags, broken links, login walls, and security tools that block legitimate bots.

For ChatGPT search specifically, OpenAI’s search crawler is called OAI-SearchBot. OpenAI’s crawler guidance covers allowing it in robots.txt and letting its published IP ranges through your firewall or security plugin.

OAI-SearchBot is separate from GPTBot, which collects data for model training. You can allow search crawling and still block training. You don’t have to treat every AI bot as one big on/off switch.

The site check in the planner further down tests both of these for you: your robots.txt rules and whether a request identifying as OAI-SearchBot gets turned away.

Have someone look at your actual configuration. Don’t paste a random robots.txt over yours, since it may contain rules you need.

It’s also worth checking Google Search Console and Bing Webmaster Tools for general indexing problems. Being indexed there doesn’t mean ChatGPT will recommend you, but if search engines can’t see your pages, that’s a problem worth fixing anyway.

Crawl access gets your content considered. It doesn’t give your plumber a reserved seat at the top of every answer.

Get your listings right

Check your website against your Google Business Profile, Bing Places, Apple Business Connect and the directories customers in your industry actually use. Fix old phone numbers, outdated hours, wrong service descriptions and stale addresses.

This one has a better reason than usual. OpenAI’s help docs say ChatGPT may share a user’s location with trusted third-party providers to get local information like nearby business listings. OpenAI doesn’t say which providers. But it’s a pretty clear signal that listing data feeds local answers, and wrong data in the wrong directory can follow you into ChatGPT.

Don’t pretend to have an office where you don’t. If you’re a service-area business, follow each platform’s rules about hiding your address and setting service areas.

Give other people something real to say about you

Your website naturally thinks you’re wonderful. Your mother probably agrees. A potential customer wants another opinion.

This is where most “rank in AI” advice gets vague, so here’s a starting point by business type. Think of these as three layers:

1. Core listings every local business needs (Google, Bing, Apple, Yelp, BBB, chamber of commerce, Nextdoor)
2. Industry platforms where your customers compare options
3. Verifiable credentials that let anyone confirm you’re legit

The manufacturer locators are the sleeper pick in that table. If you’re a certified installer for a brand, the brand’s dealer locator is an independent, authoritative page that names your business, your city and what you’re certified to do. Lots of contractors qualify and never check that their listing is correct.

Only list where you genuinely qualify. Don’t buy “AI authority” placements or fake reviews, and don’t join a neighborhood mutual admiration club.

Find your own gaps (this matters more than the table)

The table is a starting point. Your real list comes from ChatGPT itself. Run your test questions, open the Sources panel on each answer, and write down every site that cites your competitors but not you. That might be a city magazine’s “best of” list, a local news story, a niche directory you’ve never heard of, or a Reddit thread.

Those are the places worth working on first, because you’ve seen them show up for your exact searches.

I built the planner below to do the busywork. It checks whether ChatGPT’s search crawler can actually read your homepage, gives you a citation checklist for your type of business, and pulls every cited website out of a ChatGPT answer you paste in. Your checklist saves in your browser, and you can download it as a spreadsheet.

Use structured data, but don’t expect wizardry

Accurate LocalBusiness schema describes your business in a standard format: name, hours, address, service area. Google documents it, and it’s worth doing right.

Keep it consistent with what’s actually on the page, and validate it. Don’t invent awards, ratings or locations because a plugin gave you a box to fill in.

I wouldn’t pay for a promise that one schema setting guarantees ChatGPT citations. And I wouldn’t put an llms.txt file ahead of fixing a broken service page. Google has said its own AI features don’t need special files or special schema. OpenAI hasn’t said those things help either.

Measure it without grading your own homework

Before you change anything, save a baseline. Use a fresh chat, include the location, and don’t mention your business name when you’re testing whether ChatGPT finds you on its own.

For each test, write down the date, the exact question, whether search ran, which businesses came up, and which sources were cited. A fresh chat helps, but it doesn’t wipe saved memory or standardize your location, so note those too.

Recheck on a schedule and keep the bad results. Look at the sources behind each recommendation. Is a competitor answering a question you aren’t? Is an old directory feeding ChatGPT your previous phone number?

This is a diagnostic sample, not a census. Don’t turn ten answers into a claim that you “own 73.4% of AI search.” The decimal place won’t make it science.

Before you count visits, test the next step on your phone. Can someone tap to call, fill out the form, or book without fighting a broken button? A recommendation doesn’t help much if your contact form quietly died in March.

Then track what matters:

• Visits your analytics can identify as coming from ChatGPT
• Calls, form fills and bookings from those visits
• Customers who say they found you through ChatGPT
• Whether those leads turn into work you actually want

One heads-up on that first item. ChatGPT has been adding utm_source=chatgpt.com to links, which is how most analytics tools spot ChatGPT traffic. In mid-October, Search Engine Roundtable and others reported that ChatGPT appears to have stopped adding the tag and passing a referrer, so those clicks can now show up as Direct. If your ChatGPT traffic suddenly dropped this month, check that before you panic.

That’s why “How did you hear about us?” on your forms and phone calls matters more than ever. Neither method gives a perfect count. Together they get you closer.

Where I’d start this month

Pick one service you want more business for. Run a small baseline of local questions. Fix the technical and listing errors, then improve that service page with specific answers and real proof.

After that, work through your citation checklist, starting with the sources you saw ChatGPT cite for competitors. Track what happens. Expand once you have a page that works and a process you can repeat.

(If you’re wondering whether AI can just do the SEO work for you, that’s a different question, and I cover it in my article on whether AI can do SEO.)

Can a local business build a real advantage this way? I think so, especially if it answers a specific need better than anyone else in town currently does. It’s a strategy worth testing, not a promise that you’ll win every prompt.

Preferably while you’re doing something more enjoyable than refreshing ChatGPT for the seventeenth time.