Google AI Overviews for Contractors: What Changes for Local Leads

Google AI Overviews for contractors did not change what a four-truck plumbing company does on a job. They changed who the homeowner had already half-decided to call before the phone ever rang. What follows is a composite, drawn from several shops in the $1M to $5M range rather than one real company, and none of the numbers here are invented for effect.

I am walking through it as a narrative because the sequence is the lesson. The individual fixes are ordinary. The order they went in, and the reason each one mattered, is where most owners get it wrong.

Why did a water heater search stop showing this contractor in Google AI Overviews?

The owner called me because his lead volume had gone soft in a way his dashboard could not explain. Calls were down maybe fifteen percent month over month. His Google Business Profile still ranked in the top three of the map pack for "plumber" plus his city. His paid spend was flat. Nothing in the report he was looking at had moved.

So we did the least sophisticated thing available. We sat at his kitchen table and typed in the questions his customers actually type, on his phone, logged out, with location turned on.

"Plumber near me" looked normal. Ads on top, then the three-pack, then the organic results. He was where he had always been.

Then we typed "why is my water heater leaking from the bottom." The screen filled with something else entirely: a Google AI Overview, six or seven sentences long, explaining the difference between a leaking temperature and pressure relief valve, a failed drain valve, and a corroded tank that means replacement. It listed three source links. One was a national manufacturer. One was a home improvement site. One was a plumbing company two counties over that this owner had never heard of.

The annotation: that query is not a vanity keyword. It is a homeowner standing in a wet garage at 7am. It is the highest-intent moment in the entire funnel, and it happens ten to twenty minutes before anyone searches "plumber near me." The Overview had not pushed him down the page on the query he was tracking. It had inserted itself into a query he was not tracking at all, and it answered the homeowner's real question so completely that a share of those people never ran the second search.

Wondering which AI tool pays for itself first? Request a free AI audit and find out.

What do Google AI Overviews pull from for contractor searches?

Here is the honest answer, and it is less mysterious than the SEO industry wants it to be. An AI Overview is assembled from pages Google already crawls and already trusts for the topic, plus the entity data it holds about local businesses. For a trade query, that means three buckets.

• Explanatory pages on your own site. Service pages, diagnostic articles, and FAQ content that answers the specific question in plain declarative sentences. This is the bucket you control most directly, and the one most contractors have neglected the longest.
• Your Google Business Profile fields. Services list, business description, hours, service area, attributes such as emergency service, and the review corpus. These feed the local module that often sits under or beside the Overview, and they are what verify you as a real, licensed, reachable business rather than a lead-broker shell.
• Third-party corroboration. Manufacturer documentation, municipal code pages, state licensing boards, trade associations. You do not control these, but you can align with them, and agreeing with the authoritative source on a fact makes you a safer citation.

Two technical details matter enormously and almost nobody checks them. First, Google's own documentation on AI features and your website confirms that the ordinary snippet controls still apply. A page carrying a nosnippet directive, or a block wrapped in data-nosnippet, is excluded from AI Overviews. If a web developer added those to your service pages years ago to stop competitors scraping your copy, you have quietly opted yourself out of the citation entirely.

Second, Google-Extended in your robots.txt does not keep you out of AI Overviews. It governs Gemini and Vertex grounding. Several owners I have talked to added it in a defensive panic in 2024, believed they had blocked "the AI," and had in fact only removed themselves from other assistants while remaining fully eligible in Search. Check your robots.txt this week. It takes two minutes and it is the only item on this list that can be wrong in a way you cannot see from the front end.

Why did the AI Overview cite a competitor instead of a local contractor?

We pulled up the competitor's cited page. It was not beautiful. It loaded slowly. The photography was stock.

But it had one thing our owner's site did not: a page titled, almost word for word, the question the homeowner had asked, followed by an answer that named the three failure points in order, said which one is a repair and which one means replacement, and stated the rough tank lifespan range at which corrosion becomes the likely cause.

Our owner's site had a page called "Water Heater Services." It opened with a sentence about their family serving the area since 1998. It listed "repair, replacement, maintenance" as three bullet points. It answered no question at all.

The annotation: an AI summarizer cannot extract a fact from a brochure. It extracts facts from sentences shaped as subject, predicate, object. "A water heater leaking from the drain valve is usually a repair." "A tank leaking from the seam requires replacement." "Texas requires a permit for water heater replacement in most jurisdictions." Those sentences are liftable. "We pride ourselves on quality workmanship" is not liftable by anything, human or machine.

This is the single largest shift in how contractor websites earn attention, and it is not a trick. Pages that answer specific questions in specific sentences get cited. Pages that announce a service category get skipped. The skill it rewards is the one you already have, which is knowing the answer, written down.

Which pages should contractors rewrite first for AI Overviews?

He did not rebuild his website. He wrote four pages over three weeks, in the truck, dictating into his phone between calls, and had them cleaned up and posted. We picked them by reading what his dispatcher wrote in the job notes, because those notes carry the actual phrases customers use on the phone.

1. The diagnostic page. "Water heater leaking from the bottom: three causes and how to tell them apart." Three failure points, named, with the repair-or-replace call stated outright for each, plus the age at which the answer usually flips toward replacement.
2. The permit and code page. What his city requires for a water heater replacement, what the inspection covers, expansion tank and pan requirements, and what happens at resale if the prior install was never permitted. This page is close to uncopyable by an out-of-market competitor, which is exactly why it earns citations.
3. The comparison page. Tank versus tankless for his climate and his typical home size, including the descaling maintenance interval that tankless owners in hard-water areas actually face, and the gas line sizing issue that turns a simple swap into a bigger job. Honest about when tankless is the wrong answer.
4. The "what happens when you call us" page. Arrival window, diagnostic process, how the estimate is presented, who does the work, what the tech does before leaving. Process transparency is a strong trust signal for a homeowner and for a summarizer trying to describe a company in one sentence.

The annotation: notice that none of these are keyword pages. They are answer pages. Each one carries sentences an Overview can quote and a specificity a national site cannot match, because a national site does not know what his county's inspector looks for on a re-pipe.

One structural detail worth the effort: put the direct answer in the first two sentences under the heading, then explain. Burying the answer under three paragraphs of context is the most common reason a genuinely good contractor page gets passed over for a worse one.

What changes on a contractor's Google Business Profile when AI Overviews sit above the map pack?

Less than the panic suggests, and much more than the neglect suggests. The profile is still the verification layer. But details that were merely nice to have are now the details that get read aloud.

What he fixed, in one sitting at the same kitchen table:

• Services list. He had four entries. He built it out to twenty-three, using the language customers use, such as "water heater replacement," "gas line repair," and "sewer camera inspection," rather than internal category names nobody types.
• Business description. Rewritten to state, in the first sentence, what the company does and where. He put the Responsible Master Plumber license number in it. Texas plumbing licensing runs through the Texas Department of Licensing and Regulation, and the RMP number is already required on advertising. Putting it somewhere machine-readable costs nothing and gives any system trying to verify a licensed contractor a clean fact to grab.
• Hours and the emergency attribute. He advertised 24-hour service and his profile said 8 to 5. That contradiction had been live for two years, and it undercut every emergency query he wanted to win.
• Questions and answers. He seeded eight real questions his dispatcher fields weekly and answered them himself, in full sentences, with the answer first.
• Reviews, with a system behind them. Not a burst. A steady request after every completed job, which is the part most shops fumble. If you want the mechanics, we wrote up how to automate customer review requests without sounding spammy.

One thing worth knowing: Google discontinued Business Profile chat and call history on July 31, 2024. If you were quietly collecting messages there, that channel is gone, and plenty of shops never replaced it. Make sure the web form and the text line actually route somewhere a human sees within minutes.

Did the AI Overview work put more calls on the board for this contractor?

Six weeks in, his total organic sessions were up modestly. Nothing dramatic. But the calls that came in were different. The dispatcher started hearing "I read your page about the permit thing" and "you're the one who said the seam leak means a new tank." Those callers booked at a noticeably higher rate than cold calls, because they had already accepted the diagnosis before dialing, and the conversation started at scheduling rather than at persuasion.

The annotation, and the uncomfortable part: impressions on the informational queries did not turn into clicks at anything like the old rate. That is not a failure of execution. It is the structural reality of this format. Pew Research Center found that users clicked a traditional search result on 8% of visits where an AI summary appeared, compared with 15% of visits where no summary appeared, and just 1% clicked a link inside the summary itself. Fewer people click. The ones who do are further along.

Which means your measurement has to change with the format. Raw session count is now a misleading number for a contractor. Four things are worth watching instead:

• Booked jobs per organic session, not sessions. If sessions fall and booked jobs hold, you are winning, not losing.
• Branded search volume. When an Overview names you and the homeowner searches your company name an hour later, that shows up as brand demand, not as a click on the original query.
• Intake capture. Ask every caller one question about where they heard of you and make the dispatcher log it in the same field every time. Free, and more accurate than most attribution software on a four-truck shop.
• Call content. If callers start quoting your pages back to you, your answer pages are working even when the analytics look flat.

That last set only works if intake is tracked properly, which is a problem I have written about at length in the marketing attribution piece about the whiteboard that cost me an afternoon.

Does any of this change for HVAC, roofing, or electrical contractors?

The mechanics are identical. The questions are not, and the questions are the whole job. Sit with your dispatcher for one hour and write down the exact phrasing customers use, then build pages against those phrases.

HVAC. The diagnostic queries cluster around "why is my AC blowing warm air," "why is my outdoor unit iced over," and "how long does a heat pump last." The requirement queries have gotten far more valuable since the refrigerant transition, because homeowners with older R-410A systems are searching whether their unit can still be serviced and what a replacement means for their existing line set. A page that answers that honestly for your market, including when a repair is still the right call, earns citations that no national blog can match. The rest of the HVAC stack is covered in the practical AI tool stack for $1M to $5M HVAC shops.

Roofing. The money questions are about damage assessment and insurance, not about shingles. "How do I know if my roof has hail damage," "does my insurance cover a 20 year old roof," "what is a supplement." These are exactly the queries where a summarizer wants a clear, non-salesy explanation, and exactly the queries where most roofing sites offer a free inspection form and nothing else.

Electrical. "Why does my breaker keep tripping," "do I need a panel upgrade for an EV charger," "what amperage does my house need." Code cycles differ by jurisdiction, and AFCI and GFCI requirements are a moving target that national content gets wrong constantly. Being right about your own county's adopted code year is a durable advantage.

The pattern holds across every trade: the diagnostic question, the code or requirement question, the repair-versus-replace question, and the process question. Four pages per major service line, written by someone who has actually stood in the crawl space.

What should contractors stop doing because of Google AI Overviews?

Three things, stated plainly.

Stop publishing thin location pages. Twenty near-identical "Plumber in [Suburb]" pages with the town name swapped were always weak. A summarizer has nothing to extract from them, and Google's helpful content systems treat them as exactly what they are. One genuinely local page that names the inspector's requirements, the common housing stock, and the typical failure pattern in that neighborhood outperforms all twenty.

Stop adding review schema about yourself to your own site. Google's structured data guidance treats self-serving review markup as ineligible for review rich results. The star ratings that matter live on your Business Profile. Put your effort there instead. Do keep LocalBusiness and FAQPage markup accurate, because clean structured data helps a machine understand who you are even when it does not produce a visible badge.

Stop assuming the Overview is the ad unit. The Google Guaranteed badge in Local Services Ads is verified separately through license and insurance checks. It is not something an AI Overview grants or reflects. Owners conflate the two and buy the wrong fix.

What can contractors fix in 45 minutes for Google AI Overviews?

If you do nothing else this month, do this, in order:

1. Log out, turn location on, and run your ten most common customer questions as searches. Screenshot what appears. That is your actual search result, not what your rank tracker says.
2. Open your robots.txt and your page source. Look for nosnippet, data-nosnippet, and Google-Extended. Understand what each one is actually doing before you change anything.
3. Pick the single diagnostic question your dispatcher answers most often. Write a page that answers it in declarative sentences, with the repair-or-replace call stated outright in the first two sentences.
4. Fill in every empty field on your Google Business Profile, and make the hours match what you advertise.
5. Confirm that a web form submission or a missed call reaches a human within five minutes. An Overview can send you a better-informed lead. It cannot answer the phone. If that part of your operation is shaky, the speed-to-lead math is the place to start, and it matters more than anything else on this list.
6. Audit where leads are currently falling out of your intake. Being cited by an Overview and then losing the caller at the booking step is the most expensive version of this problem, and it is walked through step by step in the seven places a service business loses the job before it hits the calendar.

The uncomfortable truth in all of this is that AI Overviews reward the contractor who genuinely knows their trade and their jurisdiction, and pass over the one who bought a template site with the town name find-and-replaced. That is a fairer deal than most of what search has handed the trades over the last decade. The work is writing down what you already know, in sentences a machine can quote and a homeowner can trust. Nobody can outsource that from you, which is precisely why it holds.

At Turnkey AI we spend most of our time on the intake side of this, meaning what happens after the homeowner decides to call. But the two halves are one system. Getting cited is worth nothing if the call rings out, and a fast, clean intake is worth less if the homeowner never learns you exist. Fix whichever one is leaking harder first.

What else do contractors ask about Google AI Overviews?

Frequently asked questions

Can I opt out of Google AI Overviews without losing my organic ranking?

Not cleanly. The nosnippet and data-nosnippet controls that remove you from AI Overviews also strip your regular search snippet, which usually costs more traffic than it saves. Google-Extended in robots.txt does not affect AI Overviews at all. For nearly every contractor, the better move is to be cited rather than hidden.

How long does it take for a new answer page to show up in an AI Overview?

Expect weeks, not days. The page has to be crawled, indexed, and judged useful for the query before it becomes citation-eligible. In the composite shop described above, the first quoted pages started influencing calls around week five. Submitting the URL in Search Console speeds up crawling but not trust.

Do AI Overviews replace the Google Business Profile map pack for local searches?

No. On clear local intent queries such as plumber near me, the map pack still dominates and often appears without an Overview at all. Overviews take over the diagnostic and requirement searches that happen earlier, which is why the profile and the answer pages are two separate jobs that both need doing.

Should a contractor write these pages with AI, or by hand?

Dictate the substance yourself, then use an AI tool to clean up grammar and structure. The parts that earn citations are the local code details, the failure patterns, and the repair-or-replace judgment, and a general-purpose model does not know those for your county. Generic AI copy is exactly what gets skipped.

How many pages does a contractor actually need for AI Overviews to notice?

Four to six genuinely good answer pages per major service line beats forty thin ones. Start with the single question your dispatcher answers most often, publish it, then add one page a week. Most shops in the $1M to $5M range see the pattern working before they finish the first service line.

Does getting cited in an AI Overview send traffic, or just visibility?

Mostly visibility that converts later. Pew Research Center found only about 1% of users click a link inside an AI summary. The value shows up as branded searches, direct calls, and callers who already accept your diagnosis, which is why booked jobs per session is the metric that matters now.