22% of homeowners are now asking ChatGPT for a contractor instead of Googling one. When was the last time anyone at the shop checked if the business shows up in that answer?
Journela's April 2026 contractor marketing roundup cites Scorpion data, also referenced in ServiceTitan's 2026 industry report: nearly one in four homeowners now use ChatGPT to find a contractor. The same compilation points to the 2026 AI + Search Behavior study. 37% of consumers start service research in tools like ChatGPT or Perplexity. 73% of homeowner research for services now happens outside traditional Google Search.
That is not a future problem. That is this week's dispatch board. A leaking water heater in Paramus used to produce a Google search, three phone calls, and a job for whoever picked up first. Now a lot of those homeowners type the problem into a chat box and get three to five names. If the shop is not one of them, the truck never rolls.
What "find a contractor" looks like in ChatGPT
Homeowners do not type "plumber." They type the problem.
"Hot water heater leaking from the bottom. Who can come tonight in Passaic County and is licensed."
ChatGPT answers with a short list. Sometimes it cites sources. Often it does not. TechCrunch reported Similarweb data in July 2026 showing only 6.8% of U.S. ChatGPT desktop queries included citations as of May 2026. Travel, retail, and sports get cited more often. Local trades do not.
So the model is making a recommendation from whatever it can verify: listings, reviews, news mentions, and pages that clearly state the trade and the town. A pretty homepage that never says "licensed plumber serving Bergen County" in structured data is a brochure. It is not a source.
That is AI visibility. The question is not "does the site rank." The question is "does the answer say the shop's name."
Why ChatGPT skips a shop that ranks on Google
A shop can sit in the Google 3-Pack and still get a blank stare from ChatGPT. The engines do not share a brain.
ChatGPT still leans on Bing and on pages it can parse. Gemini leans on Google Business Profile. Google AI Overviews pull from pages Google already trusts. Ranking on Google is not the same as being recommended.
The usual misses are boring and expensive:
The shop never claimed Bing Places, so ChatGPT has no listing to trust.
The star rating sits under the floor most 2026 local-AI studies treat as a filter. A 4.1 shop with 400 reviews can still get skipped.
The website never uses LocalBusiness schema, so a machine has to guess the trade, the phone, the hours, and the towns.
Service pages are generic. "We do it all" is not a page ChatGPT can cite for a sewer backup in Essex County.
How AI engines pick local businesses is a different punch list than a classic SEO audit. Google-only work misses half of it.
The 22% is the floor, not the ceiling
22% already using ChatGPT to hire is the number that should get a shop owner's attention. The surrounding numbers are worse for anyone still treating Google as the whole funnel.
Journela also cites 37% of consumers starting service research in AI tools, and 73% of homeowner research happening off traditional Google. Younger homeowners are further along. The same roundup cites Carvium AI 2026 data: 47% of consumers use AI to assist purchase decisions, rising to 56% among Gen Z and 55% among Millennials.
Google is not standing still either. AI Overviews went from 15% of searches to 43% in a year, per Similarweb via TechCrunch. The homeowner who still opens Google often never sees the blue links.
Two doors, same job. ChatGPT names three shops. Google writes an answer box. The shops inside those answers get the call. Everyone else waits for the overflow.
The punch list that gets a shop named
Skip the theory. These are the checks that change whether ChatGPT has a reason to say the name.
Claim Bing Places and finish the listing. Hours, service area, photos, categories. ChatGPT cannot invent a recommendation from an empty record.
Get the public rating above the floor and answer the reviews. Engines treat silence as a risk signal.
Put LocalBusiness schema on the site so a machine can read name, trade, phone, hours, and towns without guessing.
Write one page per job the shop wants. Water heater replacement in Passaic County is a page. Emergency sewer backup is a page. Answer engines cite specific pages, not brand vibes.
Then measure the only number that matters. Ask ChatGPT, Gemini, and Google for the jobs the shop actually wants, in the towns it actually serves. Write down who got named.
FirstCall's free audit scores that. How to get ChatGPT to recommend a plumbing business walks the same list for plumbers. Pricing starts after the score, not before.
FAQ
Are homeowners really using ChatGPT to find contractors? Yes. Journela's 2026 stats roundup cites Scorpion data, also referenced by ServiceTitan: 22% of homeowners now use ChatGPT to find contractors.
Why doesn't ChatGPT recommend my business? It names shops it can verify. Empty Bing listings, thin reviews, and a site with no structured trade or service-area data give the model nothing to stand on.
Is Google still how most homeowners start? Less often. The same Journela compilation cites the 2026 AI + Search Behavior study: 37% start in AI tools, and 73% of homeowner service research now happens outside traditional Google Search.
Does a #1 Google ranking fix this? No. Google ranking and ChatGPT recommendations are different systems. A shop can win the map pack and still be missing from the chat answer.
What should a shop do this week? Run the names. Ask ChatGPT for the jobs and towns that pay the bills. Then run a free AI visibility audit and fix listings, schema, and reviews in that order.