Which AI Engines Actually Send Contractors Leads

The five engines are not one channel

Homeowners used to type "AC repair near me" and pick from a list. Now a chunk of them ask a question and take the answer. But "AI search" is not a single box. It is at least five different products, each with its own way of deciding which contractor to name. Treat them as one and you will optimize for the wrong thing.

Two things separate them: how many people use each, and where each one gets its facts. An engine tied to a search index (Google, Bing) reads the live web every time. A model answering from training alone is working off a frozen snapshot that may be a year stale and will never mention a shop that opened recently. Knowing which is which tells you why a given engine skips you.

The pattern: the engines with the most reach for a home-service query are the ones wired into a live search index and your Business Profile. That is good news. It means the same source pages and structured data that make you citable in one make you citable across several. You are not building five separate campaigns. You are building one clean set of facts about your shop and pointing every engine at it.

One more distinction worth keeping straight. Some of these engines answer with links you can click, and some just talk. Perplexity and, increasingly, ChatGPT show their sources, so a mention there can actually send a homeowner to your site or phone. An engine that answers from memory with no citation might describe your trade well and still never hand anyone your name. When you weigh where to spend effort, weigh both reach and whether a mention there is the kind that turns into a call.

Google AI Overviews: where your buyers already are

This is the one to win first, and it is not close. AI Overviews appear at the top of ordinary Google results. The homeowner did not choose an AI tool. They searched like always, and Google put a written answer above the blue links. That means the reach dwarfs every standalone chatbot combined, because it rides the search everyone already does.

What it pulls from is familiar: your Google Business Profile, your organic pages, and reviews. If you already rank on page one and your Business Profile is complete and consistent, you have a real shot at being named in the Overview for questions like "who repairs tankless water heaters in Tampa" or "average cost to replace a roof in Ohio." The Overview often lists two or three shops and links them. Being one of those three is the goal.

Two failure modes are common. First, a thin or inconsistent Business Profile: wrong hours, missing service list, a name that does not match your website, and the Overview has no clean entity to trust. Second, pages that answer a homeowner's question with a sales pitch instead of a straight answer. AI Overviews reward the page that plainly says "a standard tankless flush takes about an hour and runs X to Y" over the page that says "call us for the best service in town."

There is also a timing factor most shops miss. AI Overviews do not appear on every query. Google tends to show them on questions with a clear informational answer ("how long does a roof replacement take") more than on raw "plumber near me" searches, where it still leans on the map. So the pages that earn you citations are the ones that answer the homeowner's real question in plain language, not just your service list.

One boundary worth stating plainly: the map pack, the three-listing local box, and "near me" proximity ranking are a separate discipline (local SEO). AI Overviews lean on that groundwork but are not the same thing. If your question is "why am I not in the map 3-pack," that is local SEO. If it is "why does the written AI answer above the map never mention us," that is the citation and entity work this page is about.

ChatGPT: the widest standalone tool

ChatGPT is the most-used AI product a homeowner is likely to open on purpose. For a long stretch it answered purely from training data, which meant it worked off a frozen snapshot and would confidently name shops that had closed or skip ones that opened after its cutoff. That has changed. ChatGPT now searches the live web through Bing for questions that need current, local answers, and it cites the sources it used.

Two behaviors matter for a contractor. When a homeowner asks something specific and local ("licensed electricians near Fort Myers who do panel upgrades"), ChatGPT tends to run a live search and lean on what it finds: your site, your reviews, directory listings, anything that corroborates you exist and do that work. When the question is general ("how much does a panel upgrade cost"), it may answer from training and name whoever was prominent in its data, which is harder to influence quickly.

• Get your entity clean. Same business name, phone, and address everywhere the web mentions you. Mixed information gives the model a reason to skip you.
• Own a source page that answers the question directly. A page that plainly answers "what does a panel upgrade cost in Southwest Florida" is the kind of thing a live search surfaces and quotes.
• Earn third-party corroboration. One website saying you are good is a claim. A licensing board, a review platform, and a local directory agreeing is a fact the model can lean on.

There is a quirk of ChatGPT worth knowing. Ask it the same local question two ways and you can get two different answers, because whether it runs a live search depends on how the question reads. "Who should I call for a burst pipe in Sarasota tonight" pushes it to search. "Tell me about plumbing" does not. So when you test your standing there, phrase your questions the way a homeowner in a hurry actually would, with a place and a job in them. That is the query where a live search runs and your clean source page has a chance to get quoted.

Because ChatGPT reaches so many people and now reads the live web, it is the standalone engine most worth the effort after Google. The work that makes ChatGPT name you also helps Bing-fed Copilot, so you get two for one.

Perplexity, Gemini, and Copilot: the rest of the field

The other three are worth understanding, but none should jump ahead of Google and ChatGPT for a home-service shop.

Perplexity is the answer engine built around citations. Every answer shows its sources as numbered links, right there. That makes it the easiest engine to check your standing on: ask it "best gutter installers in Raleigh," and you either see your link in the sources or you do not. Its audience is smaller and skews toward researchers and heavy searchers, but the intent is high and the feedback is instant. Use it as your test bench even if its raw reach is modest.

Gemini is Google's chat assistant and pulls from the Google index, so the groundwork that wins AI Overviews carries over. Reach is growing, especially on Android phones where it is built in. For most contractors, doing the Google work well means Gemini follows without a separate push.

Copilot is Microsoft's assistant, wired into Bing and living inside Windows and Edge. Its reach for local trade queries is moderate, but here is the payoff: it reads the same Bing index that feeds ChatGPT's search. Clean up your presence for one and you have largely handled the other.

The takeaway is not that these do not matter. It is that they mostly share plumbing with the two big engines. Do the Google and ChatGPT work right and you have quietly covered four of the five.

What actually decides which contractor gets named

Reach tells you which engine to prioritize. The next question is why the engine picks one shop over another. Under the hood, none of them are guessing. They are reading a set of signals about who you are and whether you can be trusted for this exact question. Four signals do most of the work.

1. Entity clarity. The engine has to be sure which business you are. One consistent name, address, phone, and service list across your site, your Business Profile, and every directory. Contradictions make you a weak, skippable entity.
2. Structured data it can parse. Schema markup on your pages tells the model, in a format it reads cleanly, what you do, where, your hours, and your service area. It is the difference between the engine inferring your facts and being handed them.
3. Citation-worthy source pages. A page that directly answers the homeowner's question, with real numbers and specifics, is the kind of thing an engine quotes. Sales copy is not quotable. A straight answer is.
4. Third-party corroboration. AI leans hard on agreement across sources. Reviews, licensing records, and reputable local listings all saying the same thing about you is the trust an engine wants before it puts your name in a homeowner's hands.

These four signals reinforce each other. A page with clean schema and a straight answer is easy for the engine to parse, but it carries more weight when a licensing board and a review platform back up what the page claims. That is why a shop with a great website but a thin, contradictory presence everywhere else still gets skipped: one strong source cannot outvote a pile of mismatched ones. The work is less about one perfect page and more about getting every place the web mentions you to tell the same story.

Notice what is not on that list: paying for it. There is no ad slot inside an organic AI answer today. You cannot buy your way into an AI Overview citation the way you buy a Google Ad. That is the opportunity and the catch. The channel rewards the shop that did the underlying work, and it takes real time to build. For competitive terms, plan on 4 to 9 months of consistent work before you see steady mentions, not a switch you flip next week.

How to tell if any of them are naming you

You cannot fix what you cannot see. Before spending a dime on this channel, find out where you actually stand. The measurement is more manual than a rank tracker, but it is honest and you can do the first pass yourself in an afternoon.

Write down five to ten questions a real homeowner would ask before hiring your trade. Mix them: some with your city ("emergency plumber in Naples"), some general ("signs you need a new roof"), some comparison ("repair or replace a 15-year-old AC"). Then ask each engine, one at a time, and record whether your shop shows up, whether it is linked, and who got named instead.

• Perplexity first. It shows sources plainly, so it is the fastest read. If your link is in the citations, that is a real win you can see.
• ChatGPT with search on. Ask your local questions and watch whether it runs a live search and cites you.
• Google AI Overviews. Run your questions as plain Google searches and see if an Overview appears and who it names.

Two honest cautions. AI answers are not deterministic: ask the same question twice and you can get two slightly different answers with different shops named. So do not read one miss as a verdict. Run each question a few times and look at the pattern. Second, competitors being named while you are absent is the useful signal. It means the query has an AI answer and the slot exists. You are just not in it yet, which is a fixable problem, not a dead end.

If that legwork tells you the engines are quoting your competitors and skipping you, that is exactly the gap this work closes. A visibility audit turns the guesswork into a specific list: which questions, which engines, and what is missing in your entity and source pages that keeps your name out of the answer.