Is Anyone Really Asking AI for a Plumber?
A local business owner's guide to a statistic that is true, alarming, and mostly misunderstood.
All figures current and all source links verified as of publication.
If you own a business in St. Charles County, you have probably gotten the email by now. It says customers have stopped using Google. It says they ask ChatGPT instead. It usually arrives with a number attached, and the number is big.
Some of them are. Almost none of them stopped using Google.
That is the short answer. Here is the long one, with every source named, linked, and checked, because a post about checking statistics ought to be checkable itself.
The number everyone is quoting
BrightLocal runs a survey every year called the Local Consumer Review Survey. The 2026 edition surveyed a representative panel of 1,002 US adults through SurveyMonkey.
Use of ChatGPT and other generative AI tools for local recommendations went from 6% last year to 45%, making it the third most popular source of business recommendations — ahead of Yelp and Tripadvisor, behind only Google and Facebook. Over the same period, use of Google reviews slipped from 83% to 71%.
ChatGPT leads among the tools, used by 31% of consumers for business recommendations in the past year, with Google's AI Mode second at 23%.
The trust numbers move the same direction. 42% of consumers now trust AI-platform recommendations as much as written reviews. Among people who actually use AI for this, 63% trust its recommendations and only 10% distrust them.
There is a generational split worth noting. 64% of consumers aged 30–44 have asked AI for a business recommendation, against 24% of those 60 and over. If you sell to homeowners in their thirties, this is already your market.
That is a genuine shift, and it happened fast. Now read the next section.
The number nobody quotes
Surveys measure what people say. Clickstream research measures what people actually do — real browsing behavior across large device panels. It tells a calmer story.
SparkToro, working with clickstream partner Datos, found that 95% of Americans still use traditional search engines every month. That figure has moved less than 1% in two and a half years, even as AI tool adoption climbed from 8% to 38%. Rand Fishkin's conclusion was blunt: when people adopt AI tools, their Google searches go up, not down. (Worth knowing: that study is desktop-only. Fishkin says so plainly and notes it biases against AI tools, since most AI use happens on phones.)
Ahrefs ran the traffic side in February 2026, analyzing 76,000 websites. ChatGPT has reached roughly 12% of Google's search volume — a real number, not a rounding error. But Google accounts for nearly 40% of all website traffic while ChatGPT accounts for 0.21%. That is Google sending about 190 times more traffic to websites, with ChatGPT's click-through rate running roughly 96% lower.
The most careful sizing comes from Graphite.io. In March 2026, CEO Ethan Smith published an analysis combining web and mobile app usage across the five largest AI products and the six largest search engines — important, because 83% of AI usage happens in mobile apps that web-traffic-only studies miss entirely. Counting all AI sessions, AI equals 56% of global search engine volume. But when you isolate only search-style "asking" prompts and exclude the writing, coding, and brainstorming that make up most AI use, the true search-competing figure is 28% worldwide and 17% in the US.
Smith's summary is the line worth remembering: search has not decreased, and neither has Google. The pie got bigger. Combined usage of search engines plus AI grew 26% worldwide.
Smith also ends that report with a section titled "Incentives of the Author," disclosing that he runs an agency that profits from the conclusion. More research should do that.
Addition, not migration
Ask someone "have you used AI to find a local business in the past year," and 45% say yes. That is a real answer to a real question. It is not the same as "45% of the people who hired a contractor last month found him through ChatGPT."
People did not leave Google. They added a second tool. They ask an assistant, get a couple of names, then go look those names up the old way — reviews, website, phone number. BrightLocal found that 97% of AI users at least sometimes double-check AI recommendations against real reviews, and 88% check whether a cited review is legitimate or look at the source.
If you take nothing else from this post, take that. Your Google presence did not stop mattering. Something got added on top of it.
The shift that actually matters
Here is the part the panicked emails get wrong.
A large chunk of the change is happening inside Google, not outside it. Google's AI Overviews now answer the question at the top of the page. The blue links are still there. Fewer people scroll to them.
Similarweb's clickstream panel puts zero-click searches — queries that end without a click to any website — at 68% of all Google searches as of June 2026. Ten years ago it was about 45%. Pew Research, watching real browsing behavior, found people clicked a link on 8% of searches when an AI Overview was present, versus 15% when it was not. Roughly half the clicks, gone.
One caution while you are reading other people's posts on this. You will constantly see the claim that zero-click searches rose from 56% to 69% after AI Overviews launched. That figure is real, it comes from Similarweb, and it measures news-related searches between May 2024 and May 2025. Most blogs quoting it have dropped that qualifier and applied it to all searches. I nearly did too.
Either way, the direction is not in dispute, and combining it with the AI numbers gives you the real problem:
An answer layer now sits above the map pack, in both places.
You can hold a spot in Google's local three-pack and still not get named when someone asks an assistant who to call. Those are two different systems reading two different sets of signals. Winning one does not win the other.
That is the actual change. Not that customers left. That a summary now stands between your customer and your listing, and you are either in the summary or you are not.
Try this before you believe anyone, including me
You do not need to hire someone to find out where you stand. Do this yourself, tonight, for free.
Open ChatGPT, or Gemini, or Google's AI Mode. Type the question a customer would actually type. Not your business name — the need:
"Who's the best HVAC company in St. Charles, Missouri?"
"I need a tree service near O'Fallon MO. Who should I call?"
"Recommend a machine shop in the St. Louis area."
Then read the answer and ask three things.
- Are you in it? If not, who is, and what do they have that you do not?
- Is what it says about you correct? Wrong hours and dead phone numbers are common, because assistants pull from listings most owners have not touched in years.
- Where did it get that? Most tools will name their sources. That list tells you exactly which pages are shaping your reputation.
Run the same three questions in plain Google. Compare. If the two results look different, you have just found the gap this whole post is about.
This is the manual version of the Discoverability Check I built for myself — same three questions, run every month instead of once, across more prompts than you would think to try.
What I would actually do about it
Nothing here is exotic, and none of it requires a new budget line called "AI."
Fix your listing data first. Assistants pull from structured sources — your Google Business Profile, directories, aggregators. If your hours are wrong in three places, you have taught three systems to be wrong about you. This is the highest-return hour you will spend and it costs nothing. It is also most of what a website health check looks at, and you can do it yourself in an afternoon.
Write the pages that answer the question. Assistants recommend businesses they can describe. A site that says "Quality Service Since 1974" gives a machine nothing to work with. A page that says what you do, where you do it, what you charge for, and what you will not take on, gives it everything.
Get your reviews current. Recency now beats volume. In the same BrightLocal survey, 74% of consumers only care about reviews written in the last three months, and 31% will only use a business rated 4.5 stars or higher — nearly double the prior year. A separate finding: 47% won't use a business with fewer than 20 reviews. Review data feeds the summaries on both sides.
Get mentioned somewhere other than your own website. Assistants weigh third-party context heavily. A local news mention, a supplier's contractor page, an association listing, a real customer writing a real paragraph about you somewhere public. One of those is worth more than a page of your own copy claiming the same thing.
Then check it monthly. Not because it changes weekly, but because you cannot manage what you never look at.
What I am not going to tell you
I am not going to tell you that you are losing customers to ChatGPT right now. I do not know that, and neither does anyone who emails you claiming it.
What I can say is narrower and better supported. A meaningful share of your customers now see a summarized answer before they see a list of businesses. That summary is assembled from data you mostly control and mostly have not checked. Getting it right is cheap. Finding out you were wrong about it in two years is not.
Run the searches above. If you do not like what you see, that is worth an hour of your Saturday.
I run these checks monthly. If you want yours, reply with your business name and I'll send it — no call, no pitch, just what I see.
Sources
Every figure in this post, with its origin and date. All links verified August 5, 2026. Where a number is commonly misquoted, I have said so.
BrightLocal — Local Consumer Review Survey 2026 (representative panel of 1,002 US adults via SurveyMonkey)
https://www.brightlocal.com/research/local-consumer-review-survey/
Source of: 6%→45% AI usage; Google reviews 83%→71%; 74% three-month review recency; 31% requiring 4.5+ stars; 47% requiring 20+ reviews.
BrightLocal — "Nearly Half of Consumers are Asking AI for Business Recommendations" (March 10, 2026)
https://www.brightlocal.com/research/lcrs-ai-trust/
Source of: 42% trusting AI as much as written reviews; ChatGPT 31% and Google AI Mode 23%; the 64% / 24% age split; 63% trust and 10% distrust among AI users; 88% checking sources and 97% double-checking against reviews.
Note: BrightLocal states that some figures in this report are drawn only from the 455 respondents who used an AI tool in the last 12 months. Where a figure applies to AI users rather than all consumers, I have said so above and in the text.
Graphite.io — "AI Is Much Bigger Than You Think," Ethan Smith (March 2026)
https://graphite.io/five-percent/research/ai-is-much-bigger-than-you-think
Source of: AI at 56% of global search volume; 28% worldwide and 17% US for search-style prompts; 26% growth in combined usage; 83% mobile app usage.
Ahrefs — "ChatGPT Has 12% of Google's Search Volume," Patrick Stox (February 11, 2026; Ahrefs Web Analytics across 76,000 websites)
https://ahrefs.com/blog/chatgpt-has-12-percent-of-googles-search-volume/
Source of: 12% search volume share; 190× traffic gap; 39.98% vs 0.21% traffic share; 96% lower CTR.
SparkToro / Datos — "20% of Americans use AI tools 10X+/month," Rand Fishkin (August 26, 2025; multi-million device clickstream panel, US, Jan 2023–Jan 2025)
Source of: 95% monthly traditional search usage; AI adoption 8%→38%.
Note: often misquoted as "95% of ChatGPT users also use Google." The actual finding is that 95% of Americans use traditional search engines monthly. The study is also desktop-only, which its author notes biases against AI tools.
Similarweb — "Zero-Click Marketing: What the 2026 Data Means," Limor Barenholtz (June 10, 2026)
https://www.similarweb.com/blog/marketing/geo/zero-click-marketing/
Source of: 68% zero-click rate across all Google searches; ~45% a decade ago.
Similarweb data reported by Digiday, Sara Guaglione (July 10, 2025)
Source of: zero-click rise from 56% to 69%, May 2024 to May 2025.
Note: this measures news-related searches specifically. It is very widely quoted as applying to all Google searches. It does not.
Pew Research Center (July 22, 2025)
Source of: 8% click-through with an AI Overview present versus 15% without.
A note on one figure I left out. SparkToro's widely cited finding that Google receives ~373× more searches than ChatGPT is real, but it reflects 2024 data published in March 2025. Ahrefs' 2026 measurement puts ChatGPT at roughly 12% of Google's search volume. Both used defensible methods; they are measuring different years and different things. Quoting the 2024 ratio as current would understate how much has changed, so I have used the newer figures instead.
A note on who paid for this research. BrightLocal sells local SEO software. Ahrefs, Similarweb, and SparkToro sell SEO and analytics tools. Graphite is a growth agency. Every source above has a commercial interest in you caring about this topic. That does not make the data wrong — the methodologies are disclosed and the panels are large — but it is why I went to the original reports rather than to the blog posts summarizing them, and why two of the numbers in this post are corrections to what those summaries say.
I build and maintain websites for small businesses around the St. Louis metro, and this is part of what that includes.