AI search has changed which searches send you traffic, not whether search still works. AI Overviews now dominate the research questions people ask before they hire anyone, and they appear on only about 15% of local-intent searches like "plumber near me". For a local service business, the map pack still runs the show.
Key takeaways
- Local-intent searches still show the map pack 93% of the time and an AI Overview only 15% of the time. Informational searches flip that almost exactly.
- The scary "34.5% click drop" figure everyone quotes was measured on informational searches only. It does not describe local search, and it gets misapplied constantly.
- 45% of people now use AI tools for local recommendations, up from 6% a year earlier. But 97% of them still check the answer against real reviews before acting.
- AI cites third-party sites more than your own. Around 60% of citations for local service queries went to Yelp, Reddit, Thumbtack and similar, not to business websites.
- The real thing to watch is Google's AI Local Pack test, which cut the map pack to one or two listings and, where it appeared, cost some businesses over half their calls.
Nearly everything written about AI and small business SEO is written for marketers, and it treats "AI search" as one undifferentiated wave coming for your traffic. That's not what the data shows for a service business. What's happened is more specific and more useful than that, so here's the version for someone who owns a trade business and wants to know whether to do anything differently.
The split that explains almost everything
Local search analysts at Whitespark ran 540 queries across three US cities and six service industries, including plumbers, dentists and clinics, and sorted them by what the searcher was actually trying to do. The results split cleanly.
| What someone searched | AI Overview appeared | Map pack appeared |
|---|---|---|
| Local intent, e.g. "plumber near me" | 15% | 93% |
| Informational, e.g. "how much does a water heater cost" | 92% | 6% |
| Hybrid, e.g. "average cost of dental implants in Phoenix" | 97% | 17% |
A second, larger study by Local Falcon covering 60,000 queries found a lower overall rate, around 40% of local business searches triggering an AI Overview against Whitespark's 68%. Different query mixes produce different headline numbers, so treat the overall figure as a range. The pattern inside it held in both: the closer a search gets to "I want to hire someone now", the less AI is in the way.
That's the whole story in one line. AI took the research phase. It has not taken the hiring phase.
The statistic you should stop being frightened by
You've probably seen a version of "AI Overviews cut clicks by 34.5%". It comes from a real study by Ahrefs across 300,000 keywords, and it's solid work. It also explicitly excluded commercial, navigational and local queries. Over 99% of what it measured was informational.
So it's an accurate description of what happened to blog traffic and a poor description of what happened to a plumber. Nobody has yet published a rigorous study isolating click-through loss for local service searches specifically, and that gap is worth knowing about, because plenty of agencies quote the 34.5% figure at local businesses to sell an urgent fix.
What is true: if a chunk of your traffic came from general how-to content, you've lost some of it and you'll lose more. If your traffic came from people in your town searching for your service, much less has changed than the headlines suggest.
People really are asking AI for recommendations now
This part is real and it moved fast. BrightLocal's 2026 consumer survey of just over a thousand US adults found 45% had used an AI tool to get local business recommendations, up from 6% the year before. That puts AI third behind Google and Facebook as a discovery route, ahead of Yelp and TripAdvisor.
Among those users, ChatGPT led at 31% and Google's AI Mode was second at 23%. Adoption skews younger, peaking at 64% in the 30 to 44 group and dropping to 24% among the over-60s, which matters if your customers skew older.
The number that keeps this in proportion, from the same survey: 97% of people who get an AI recommendation cross-check it against actual review platforms before doing anything about it. So AI is increasingly the thing that produces a shortlist, and your reviews are still the thing that wins the job off that shortlist. That has a very practical consequence, covered below.
What AI actually cites when someone asks for a plumber
Here's the finding that should change where you spend your time. When researchers looked at what AI systems cited for local service queries, roughly 60% of sources were third-party pages, Yelp, Reddit, Quora, Thumbtack, HomeGuide, even job boards. Only about 40% were individual business websites.
Yelp in particular dominates. One analysis of tens of millions of AI responses found Yelp receiving 3.4 times more citations than its nearest competitor, with "near me" style queries driving 290,000 Yelp citations against 87,000 for every rival combined. Two thirds of those came through Google's AI Mode specifically.
Google's own AI Mode, according to Whitespark's 2026 guide to it, pulls from Google Business Profiles, review platforms with Yelp named explicitly, local news and community sites, and social content. Click a map marker inside AI Mode and you get a card with ratings, reviews, photos, hours and a call button, which is a Google Business Profile in everything but name.
Read those together and the conclusion is uncomfortable but simple: your presence in AI answers is built mostly outside your own website. Your Google Business Profile, your review count and quality, and your presence on the directories that AI leans on do more work than any page you write.
If your Google Business Profile is thin or your review flow has stalled, that's the highest-leverage thing to fix before worrying about AI at all.
See local SEO for service businessesThe AI Local Pack, and why it's the real thing to watch
Everything above is broadly reassuring for a local business. This part isn't.
Through late 2025 and into 2026, Google has been testing an "AI Local Pack" that replaces the familiar three-business map pack with one or two AI-selected businesses, in some versions without a direct call button. Local SEO specialist Joy Hawkins tracked it and reported that some businesses saw calls to their Google Business Profile drop by more than half where the format appeared. At the time of that January 2026 report it was showing on roughly 8% of keywords and climbing, and clicks on ads rose even as organic local clicks fell.
A follow-up in August 2026 noted the format was pulling recommendations from low-quality listicles and that Google had begun adding AI-generated images to local AI Overviews. It's still an active test, not a finished rollout.
There's no clever defence against this. If the pack shrinks from three slots to one, being fourth is the same as being fortieth. The only real hedge is to be genuinely one of the strongest one or two businesses in your area on the signals Google uses, and to not have the phone be your only route in. A working quote form, a text option and an email address stop a format change from taking your whole lead flow with it.
Does schema markup get you cited by AI?
You'll be sold this one, so it's worth knowing what the evidence actually says.
A 2026 research paper analysed 730 AI citations across ChatGPT and Gemini for 75 commercial queries. Generic schema, the basic LocalBusiness or Organization markup most sites have, showed no measurable citation advantage at all. Schema with genuinely populated values, real prices and real aggregate review counts rather than empty scaffolding, was cited 61.7% of the time against 41.6% without.
And the strongest predictor of being cited, by a distance, was plain organic ranking. Pages ranking first were cited 43% of the time. Pages ranking seventh, 5%.
Separately, when Whitespark surveyed 47 local SEO experts and scored 187 factors for AI visibility, structured data landed at number 111. The factors at the top were all Google Business Profile signals: whether your profile shows as busy, how many photos users have uploaded, click-through rate from the map pack, your service categories, and reviews with photos attached.
So: add schema, because it's cheap, correct, and it helps a bit when it carries real data. Don't buy a package that sells it as an AI shortcut.
What to actually do, in order
Nothing here is new, which is rather the point. The AI era rewards the same things local SEO always rewarded, just more sharply.
- Get your Google Business Profile genuinely complete. Every category, real hours, real photos, service areas, services listed. It feeds the map pack and it feeds AI Mode. My guide to ranking a home service business on Google Maps covers this properly.
- Build review volume deliberately. Reviews decide the shortlist AI hands people, and they decide what happens after. Ask every satisfied customer, every time, with a link.
- Claim and fill in the third-party listings AI actually cites. Yelp above all, plus the trade directories in your area. You don't have to advertise on them. You do need an accurate, reviewed, complete profile there.
- Answer the research questions on your own site. Costs, timelines, what goes wrong, how to choose. You'll get fewer clicks than you used to, and this is what gets you quoted when someone asks an AI about your trade.
- Add a second and third way to contact you. Quote form, text, email. Insurance against a format change you don't control.
- Put real numbers in your schema. Actual prices and actual review counts, not empty fields.
If none of that is in place yet, start with the fundamentals in my practical local SEO guide for service businesses before layering anything AI-specific on top.
What to ignore
A short list, because the noise here is expensive.
Ignore anyone selling "GEO" or "AI optimisation" as a separate service with its own retainer, when what they describe is local SEO with a new label. Ignore the 34.5% click-drop figure applied to local businesses. Ignore claims that any agency can guarantee you a spot in an AI Overview, because nobody controls that. And ignore advice to abandon your website for social, since the AI systems in question mostly cite web pages and business listings.
Also worth resisting: rewriting all your existing content into an "AI-friendly" format. The one study that measured this found ordinary ranking mattered far more than markup or formatting tricks. If your basic SEO fundamentals aren't in place, that's where the return is.
The bottom line
For a local service business in 2026, AI search has changed the top of the funnel and left the bottom of it broadly intact. You'll get fewer visits from people researching. You'll get roughly the same volume from people ready to hire, because those searches still return a map pack.
What genuinely deserves your attention: your Google Business Profile, your reviews, your presence on the third-party sites AI leans on, and the AI Local Pack test, which is the one development that could actually cost you calls. Everything else is a variation on doing local SEO well, which was true before any of this and is more true now.
If you want a straight assessment of where your business stands on the signals that matter, send me your details and I'll take a look at your profile and your site.
Frequently asked questions
Do AI Overviews show up when someone searches for a local business?
Much less often than you'd think. A study of 540 local queries across six service industries found AI Overviews appeared on only 15% of local-intent searches like "plumber near me", while the map pack appeared on 93%. On informational searches such as "how much does a water heater cost", AI Overviews appeared 92% of the time. The searches that get you hired are still map-pack searches.
Is AI search killing website traffic for small businesses?
It's hitting informational traffic hard and local traffic much less. The widely quoted 34.5% click-through drop comes from a study that deliberately excluded local and commercial searches, so it doesn't describe what happens to a plumber or an electrician. If most of your traffic came from blog posts answering general questions, expect real losses. If it came from people searching for your service in your town, far less has changed.
How many people use ChatGPT to find local businesses?
In BrightLocal's 2026 consumer survey, 45% of people said they'd used an AI tool for local business recommendations, up from 6% the year before. That makes AI the third most-used discovery route behind Google and Facebook. The important caveat from the same survey: 97% of those people still cross-check what the AI told them against real review platforms before they act on it.
What do AI systems use to recommend a local business?
Mostly other people's pages, not yours. In one analysis of local service queries, around 60% of the sources cited were third-party sites like Yelp, Reddit, Thumbtack and HomeGuide, with only 40% pointing at individual business websites. Google's AI Mode also pulls directly from Google Business Profiles and review platforms. Your reviews and your listings do more work here than your homepage does.
Does schema markup help you get cited by AI?
Only a specific kind. A 2026 study of 730 AI citations found generic LocalBusiness or Organization markup gave no measurable advantage. Schema with real populated values, actual prices and genuine review counts, was cited 61.7% of the time versus 41.6% without. The strongest predictor by far was still ordinary organic ranking. Add schema because it's cheap and correct, not because it's a shortcut.
What is the AI Local Pack?
It's a format Google has been testing that replaces the usual three-business map pack with one or two AI-selected businesses, sometimes without a direct call button. Tracking reported in early 2026 found some businesses losing more than half their calls when it appeared, on roughly 8% of keywords at the time. It's the single development in AI search most worth watching if you rely on the phone ringing.