The new path: from question to booked job
The old path was a search box and a click. A homeowner typed "water heater repair near me," scanned the top few results, and called one. The new path has more steps, and most of them happen before your website ever loads.
It usually starts with a plain-English question in ChatGPT, Gemini, or Perplexity, or with a question typed into Google that returns an AI Overview at the top of the page. The homeowner is not searching for a link. They are asking for advice: "is a 15-year-old furnace worth repairing or should I replace it," "what does a standing-seam metal roof actually cost," "how do I know if a plumber is licensed in Florida." The AI answers in a paragraph, and somewhere in that trip it starts naming companies.
Then the homeowner narrows. "OK, who does this in Cape Coral," or "give me three licensed roofers near me who handle tile." The answer engine responds with a short list of names, sometimes with a sentence about each and a link. That short list is the new shortlist. The homeowner did not build it. The machine did, from whatever the web already says about the companies in that trade and town.
Only then does a click happen, and it lands on the sites of the two or three named companies. The homeowner checks the phone number, looks for proof you do this exact job, and calls. So by the time your website matters, the field has already been cut. You either made the AI's short list or you did not, and the deciding work happened on pages and sources you may never have thought about.
The practical point for an owner: the moment that decides whether you get the call has moved earlier, off your website and into the answer. Your site still closes the job, but it cannot close a job that never reaches it. The question is no longer only "is my site good." It is "does the AI know my shop exists, know what I do, and trust me enough to say my name."
What homeowners actually type into the AI
Homeowners do not talk to ChatGPT the way they talked to a search box. Search was two or three keywords. AI questions are full sentences, often with the homeowner's whole situation baked in. Understanding the shape of these questions tells you what the answer engine is trying to match you against.
The questions cluster into a few kinds, and they map to where the homeowner is in the trip:
- Diagnosis: "my AC is blowing warm air and the outside unit is humming, what is wrong" or "brown water stain spreading on my ceiling after rain, how bad is it." No company gets named yet. The AI is teaching.
- Decision: "repair or replace a 12-year-old water heater" or "is a metal roof worth the extra cost in a hurricane zone." Here the AI starts referencing the kind of company that handles the job.
- Selection: "who are the best-reviewed licensed roofers in Naples for tile" or "give me three plumbers near me that do repiping and are insured." This is where names appear.
- Verification: "is [company name] licensed and legit" or "what do people say about [company]." The AI is now checking a name the homeowner already has, often one it gave them a step earlier.
Notice how specific the selection and verification questions are. They carry the trade (tile roofing, repiping), the town, and a qualifier (licensed, insured, best-reviewed, emergency). The answer engine has to match all of that to a company, and it can only match what the web already states plainly. If your pages never say "we repipe homes in Naples" in words a machine can read, you will not surface for the repiping question no matter how many of those jobs you actually do.
This is why the fix is never a clever prompt or a trick. It is making the plain facts of your business (trade, sub-trades, service area, license status, the specific jobs you handle) exist as clear, quotable statements the AI can find and lift. When those facts are absent or buried, the machine fills the gap with your competitors, who wrote theirs down.
How the AI decides which contractors to name
An answer engine is not choosing a favorite. It is assembling an answer from sources it can find, parse, and trust, then reading out the names that appear across enough of those sources to feel safe. Break that into the parts an owner can influence and it stops feeling like a black box.
Find. The AI can only name a company it can locate on the web. That means a real, indexable website with pages that state what you do, plus a presence in the places the machine reads about local businesses. A shop with a thin one-page site and little else is nearly invisible to it, because there is almost nothing to find.
Parse. Finding your page is not enough. The engine has to understand it. Clear headings, plain sentences that name the trade and town, and structured data (schema) that spells out your business, services, and area in a format machines read give the AI clean facts to lift. Vague copy like "quality solutions for your home" gives it nothing to work with, so it moves on to a page that talks straight.
Trust. This is the part contractors underestimate. Before an AI reads your name to a homeowner, it wants corroboration: does the rest of the web agree this company is real and does this work. That agreement comes from your details matching across the sources it reads, from third parties mentioning you by name in your trade and town, and from your own site being consistent about who you are. When the web tells one clear story about your shop, the AI repeats it. When the story is fragmented or contradictory, the AI hedges by naming someone else.
Put simply, getting named is a function of entity clarity: does the machine understand your business as a specific, corroborated thing it can confidently point a homeowner to. That is the exact work this silo owns, and it is separate from where you sit on a page of Google links or in the map. You can rank on Google and still be missing from the answer, because ranking and being cited are not the same signal.
Why the shift matters more for contractors than most businesses
Every business is affected by AI search, but a contractor is exposed in a specific way, and it cuts both directions. The exposure is worth understanding because it changes how much this is worth to you.
First, the answer is short. A page of blue links had room for ten names; a homeowner scrolling could still spot you at number six or seven and click. An AI answer names two or three companies and stops. There is no number six in a paragraph. Either you are one of the handful the machine trusts, or you are left off the page the homeowner ever sees. The penalty for being absent went from "lower on the list" to "not on the list."
Second, contractor jobs still require a human. An answer engine can tell a homeowner how a mini-split works or roughly what a re-roof costs, but it cannot clear the drain, replace the panel, or reseal the flat roof. So the homeowner who reaches the selection question still has to pick a real company and call. The high-value moment, the one right before the phone rings, is exactly the moment the AI is now shaping. That is why being named matters so much for trades specifically: your buyers cannot finish the job inside the chat, so they always exit to a human, and the AI decides which humans they consider.
Third, the searches skew toward hiring. A homeowner asking an AI about a burst pipe or a dead AC in July is not doing idle research. They have a problem at a real address and they are close to spending money. The questions that carry a trade, a town, and an urgency word are the ones most likely to produce named companies, and they are the ones attached to real jobs. When the AI answers those with someone else's name, you did not lose a curious reader. You lost a booking.
The upside of all this: the channel is still young enough that most of your competitors have done nothing about it. The contractor who makes his business legible to answer engines now gets named while the rest are still arguing about whether any of this is real. That window does not stay open forever.
Blue-link rankings vs. getting named in the answer
Owners keep assuming that if they rank on Google, the AI will name them, and that if they are missing from the answer, the fix is more ranking work. Both assumptions are half right, which is worse than being wrong, because it sends money to the wrong place. Here is the honest split.
| The homeowner's search | Ranking on Google gets you | Being named in the AI answer gets you |
|---|---|---|
| Types keywords, scans links | A spot on the results page they scroll | Nothing extra here; this is the classic surface |
| Asks a full question in ChatGPT or Perplexity | No guarantee at all; the AI may never touch your page | Your name spoken as part of the answer |
| Reads Google's AI Overview above the links | Helps, but the Overview cites its own short source set | A citation inside the answer, above the ten links |
| Asks the AI for a shortlist of companies | Does not directly produce the list | A slot on the two-or-three-name shortlist |
The overlap is real: the pages that rank well (fast, specific, clearly written, well corroborated) are often the pages an answer engine trusts, so good SEO makes you a stronger candidate. But the overlap is not total. An answer engine weighs entity clarity and third-party corroboration more heavily than a link ranking does, and it can name a company that sits below you on Google while skipping you, because that company's business facts are cleaner and its web story more consistent. That is why you can be page-one on Google and still hear "I asked ChatGPT and it never mentioned you."
The takeaway is not "ranking does not matter." It matters, and where you sit on the page of links and in the map is its own work in its own lane. The takeaway is that being named in AI answers is a distinct outcome with distinct inputs. If you are missing from the answer, the fix lives in the citation and entity layer, not in chasing one more keyword position. Diagnosing which layer is actually failing is the first thing an audit sorts out.
What a contractor should actually do about it
Skip the panic and skip anyone selling a magic "get mentioned by ChatGPT" button. Being named in AI answers comes from making your business legible and corroborated on the open web. That is concrete work, and it goes in a sensible order.
- Find out if the AI knows you exist. Ask the answer engines the questions your homeowners ask ("best licensed [trade] in [town]," "who does [specific job] near me") and see whether you are named, named wrong, or absent. That is your real starting position, and it is usually not what owners assume.
- State your facts plainly on your own site. Every trade you run, every sub-trade, your exact service area town by town, your license and insurance status, written as clear sentences a machine can lift. A page that says "we repipe and repair water heaters in Naples and Bonita Springs" can be cited; "quality plumbing solutions" cannot.
- Add the structured data that machines read. Schema that spells out your business, services, and area gives the answer engine clean, unambiguous facts instead of asking it to guess from prose. This is the schema work that specifically decides whether an AI parses you correctly.
- Make the web tell one story about you. Consistent business details and real third-party mentions in your trade and town are the corroboration the AI trusts before it says your name. When the story is fragmented, the machine hedges to a competitor.
- Track it over time. Whether ChatGPT or Perplexity names you is not something a Google rank report tells you. It has to be checked directly and watched as it moves. Competitive positions take 4-9 months to earn, in the answer just as on the page.
None of that is a trick, and none of it is instant. It is the entity, schema, and corroboration work that decides which contractor an answer engine trusts. The shops that do it now get named while their competitors are still deciding whether AI search is real. An audit is where it starts: it tells you exactly where you stand in the answers today, delivered in 1-3 business days, before anyone quotes you a thing.