Does ChatGPT name the contractors at the top of Google Maps?
Sometimes, and less often than the top of the map suggests. We checked 8 home-service trades in 50 US metros. ChatGPT (the OpenAI API with web search) named 352 of the 1,183 businesses in the Google Maps top three. It named more from Google's local pack: 569 of 1,193. The map, the local pack and AI answers are three different lists, so it pays to check your business on all three.
By Youssef Hodaigui, Mindflow Marketing · Data collected October 3–17, 2026
- 1
A homeowner has a job to do
emergency plumber near me - 2
They look in three places
Google resultsThe mapAn AI answer - 3
They check before they call
Reviews, photos of real work and a page that answers their question.
- 4
They request an estimate
Your form or phone number, working on their phone.
Not an actual ranking, listing or result.
| Maps top three named by ChatGPT | 352 of 1,183 |
|---|---|
| Maps top three also in the local pack | 431 of 1,184 |
| Local pack named by ChatGPT | 569 of 1,193 |
| On all three lists | 205 businesses |
How did we check?
For each of 8 trades in each of 50 metros, we collected three lists on the same day. The first was the top three Google Maps listings for the trade at a fixed map point in the city. The second was the three businesses in Google's local pack for the service plus the city. The third was every business ChatGPT named in five separate answers to one homeowner question. That's 400 trade-and-city checks in all.
The question we asked ChatGPT was: "Which [trade] companies in [City], [State] should a homeowner call? Name the companies and the sources." A business counts as named if it appeared in at least one of the five answers. The full rules for collecting, matching and counting are on the methodology page.
Does it depend on your trade?
Yes, it varies. ChatGPT named 57 of the 149 plumbing businesses at the top of the map and 28 of the 146 in remodeling. In HVAC, it named 95 of the 150 businesses in the local pack but only 40 of the 149 at the top of the map. Each trade below covers the same 50 metros, and each row keeps its own base.
| Trade | Maps top three named by ChatGPT | Maps top three also in the local pack | Local pack named by ChatGPT | On all three lists |
|---|---|---|---|---|
| Plumbing | 57 of 149 | 57 of 149 | 83 of 150 | 31 |
| Electrical | 50 of 147 | 45 of 147 | 85 of 150 | 26 |
| Garage door | 46 of 146 | 67 of 147 | 66 of 150 | 29 |
| Roofing | 45 of 147 | 59 of 147 | 75 of 150 | 29 |
| Water damage | 45 of 149 | 46 of 149 | 46 of 148 | 16 |
| Pest control | 41 of 150 | 56 of 150 | 83 of 147 | 30 |
| HVAC | 40 of 149 | 56 of 149 | 95 of 150 | 32 |
| Remodeling | 28 of 146 | 45 of 146 | 36 of 148 | 12 |
| All eight trades | 352 of 1,183 | 431 of 1,184 | 569 of 1,193 | 205 |
Denominators fall short of 150 where a written ruling removed a listing: for example, a listing in the wrong trade or outside the metro. Every ruling is written down, and each check keeps the rulings in force when it closed. A smaller check of 7 more trades in 5 metros is reported separately on the methodology page and isn't added to these totals.
Is the local pack a better guide than the map?
Somewhat, but it isn't a stand-in either. Businesses in the local pack were named more often than businesses at the top of the map: 569 of 1,193 against 352 of 1,183. Yet only 431 of 1,184 Maps leaders were in the local pack at all, and just 205 businesses made all three lists. So a top spot on one Google list doesn't predict a spot on the other, or in ChatGPT's answers.
Results for each of the 50 metros
Each row adds up the 8 trades in that metro, so most denominators are 24. "Unresolved" means some named businesses couldn't be matched to a single listing; they're left out of the count rather than guessed. Detroit's Maps figures leave out 8 listings across the border in Windsor, Ontario. Each metro name links to our page for that city.
| Metro | Maps top three named by ChatGPT | Maps top three also in the local pack | Local pack named by ChatGPT | On all three lists |
|---|---|---|---|---|
| Atlanta | 5 of 24, 2 unresolved | 11 of 24 | 9 of 24, 1 unresolved | 5 |
| Augusta | 10 of 23 | 4 of 23 | 17 of 24 | 4 |
| Austin | 2 of 24 | 7 of 24 | 10 of 24 | 2 |
| Baltimore | 6 of 24 | 9 of 24 | 13 of 24 | 5 |
| Boston | 3 of 24 | 10 of 24 | 4 of 24 | 3 |
| Charleston | 9 of 24 | 10 of 24 | 9 of 24 | 3 |
| Charlotte | 9 of 24 | 10 of 24 | 9 of 24 | 3 |
| Chicago | 3 of 24 | 6 of 24 | 9 of 24 | 0 |
| Cincinnati | 6 of 23 | 6 of 23 | 12 of 24 | 2 |
| Cleveland | 2 of 23 | 10 of 23 | 4 of 24 | 1 |
| Columbia | 10 of 24 | 9 of 24 | 15 of 24 | 5 |
| Columbus | 7 of 23 | 8 of 23 | 15 of 24 | 4 |
| Dallas | 5 of 24 | 10 of 24 | 10 of 24 | 4 |
| Denver | 7 of 24 | 12 of 24 | 8 of 24 | 5 |
| Detroit | 3 of 16 | 8 of 16 | 8 of 24 | 3 |
| Durham | 12 of 24 | 10 of 24 | 17 of 24 | 8 |
| Fort Worth | 7 of 24 | 9 of 24 | 10 of 24 | 6 |
| Greenville | 13 of 24 | 8 of 24 | 18 of 24 | 5 |
| Houston | 1 of 24 | 3 of 24 | 4 of 24 | 0 |
| Indianapolis | 8 of 24 | 8 of 24 | 13 of 24 | 3 |
| Inland Empire (Riverside) | 9 of 24 | 12 of 24 | 13 of 24 | 6 |
| Jacksonville | 7 of 24 | 6 of 24 | 15 of 24 | 4 |
| Kansas City | 8 of 24 | 5 of 24 | 13 of 24 | 2 |
| Knoxville | 14 of 24 | 9 of 24 | 20 of 24 | 8 |
| Las Vegas | 11 of 24 | 7 of 24 | 13 of 24 | 4 |
| Los Angeles | 4 of 24 | 10 of 24 | 7 of 24 | 4 |
| Miami | 3 of 24 | 5 of 24 | 4 of 24 | 0 |
| Minneapolis | 7 of 24 | 10 of 24 | 14 of 24 | 6 |
| Nashville | 7 of 24, 1 unresolved | 9 of 24 | 14 of 21, 2 unresolved | 5 |
| New York | 5 of 24 | 5 of 24 | 8 of 24 | 2 |
| Orlando | 7 of 24 | 11 of 24 | 11 of 24, 1 unresolved | 3 |
| Philadelphia | 8 of 24 | 11 of 24 | 11 of 24 | 6 |
| Phoenix | 2 of 24 | 6 of 24 | 11 of 24 | 2 |
| Pittsburgh | 12 of 23 | 11 of 23 | 12 of 24 | 7 |
| Portland | 7 of 24 | 7 of 24 | 12 of 24 | 4 |
| Providence | 12 of 24 | 11 of 24 | 12 of 22 | 9 |
| Raleigh | 6 of 24 | 8 of 24 | 14 of 24 | 2 |
| Richmond | 8 of 23 | 7 of 23 | 12 of 24 | 4 |
| Sacramento | 6 of 24 | 9 of 24 | 13 of 24 | 4 |
| Salt Lake City | 9 of 24 | 12 of 24 | 12 of 24 | 7 |
| San Antonio | 2 of 24 | 7 of 24 | 6 of 23 | 2 |
| San Diego | 2 of 24 | 8 of 24 | 6 of 24 | 1 |
| San Francisco–Oakland | 8 of 24 | 12 of 24 | 15 of 24 | 7 |
| San Jose | 8 of 24 | 8 of 24 | 12 of 24 | 5 |
| Savannah | 13 of 23 | 13 of 24 | 18 of 24 | 9 |
| Seattle | 9 of 24 | 11 of 24 | 10 of 24 | 5 |
| St. Louis | 7 of 24 | 6 of 24 | 14 of 24 | 3 |
| Tampa | 8 of 24 | 11 of 24 | 10 of 24 | 6 |
| Virginia Beach | 6 of 22 | 7 of 22 | 13 of 23 | 3 |
| Washington, DC | 9 of 24 | 9 of 24 | 10 of 24 | 4 |
What does this mean if you run a home-service company?
A top spot on the map doesn't mean AI answers will name you, and a strong AI showing doesn't mean you'll rank on the map. Check all three for your own trade and town.
When ChatGPT checks a named business, it most often cites BBB, government and licensing sites, and the company's own site; we counted those sources in what ChatGPT checks before recommending a contractor. Keeping those details accurate and consistent is part of our AI search optimization work. No one can promise a business a place on the map or in an AI answer.
What can't this study tell you?
It can't tell you why a business was named, how the ChatGPT app behaves, or what happens next month. These are counts from one collection window in 50 chosen metros, and they come with these limits:
- Not random.We chose the 50 metros; they don't represent every US market.
- Not the app."ChatGPT" here means the OpenAI API with web search (model `gpt-5-2025-08-07`). The ChatGPT app may answer differently.
- Five answers per check."Named" means named at least once in five; it isn't a stable ranking.
- One snapshot.Results were collected October 3–17, 2026, and both Google and AI answers change.
- Association, not cause.The counts don't show why a business was or wasn't named.
- Rules changed during collection.Each check keeps the matching rules in force when it closed, so comparisons across metros mix rule versions.
- No names.We report totals only and don't identify any business in the sample.
Can you use or cite this data?
Yes. Download the trade-and-metro counts (no business names or identifiers) as CSV files: counts for each trade in each metro and counts for each metro. You can reuse them with credit and a link to this page (CC BY 4.0).
Suggested citation: Mindflow Marketing, "Home-Service Visibility Study (October 2026)," mindflowmarketing.com/benchmark/, data collected October 3–17, 2026.
Spotted an error? Email youssef@mindflowmarketing.com. Our editorial standards explain how we correct pages.
Questions about the study
Is this the ChatGPT app?
No. We used the OpenAI API with web search turned on, the way a developer would. The app can use other settings and context, so its answers may differ. We report what the API returned and say so beside every figure.
Why counts and not percentages?
Some denominators changed when a ruling removed a listing, so a percentage would hide what was counted. We give both numbers, "N of D", so you can see the base. It also stops small samples, like one city's 24 listings, from looking more precise than they are.
Did you include your own clients?
No. Mindflow's clients and prospects were excluded by name before collection, so none of them is in these counts. The study describes other businesses, which is also why we report totals only and never name one.
Does being on all three lists get you more calls?
We didn't measure calls or booked jobs. The study counts which lists a business appeared on, not what customers did after seeing it. That would need call and booking records, which this study didn't use.
Get your free AI Visibility Review
See how your business shows up in your market on Google, on the map and in AI answers. Fill in a short form. We'll email you a written report with a link to book a 20-minute walkthrough. You'll have it within 2 business days. It shows where you appear on the map for three core searches in your area, the gaps in your Google Business Profile, how your reviews compare with your top three competitors, what Google's AI and ChatGPT say for two searches, and the three fixes we'd make first.
No obligation. If we're not the right fit, we'll tell you on the walkthrough.