45 percent of customers now use ChatGPT to research a local business. ChatGPT recommends 1.2 percent of them.

ChatGPT is now the third most popular way customers find a local business

Two studies landed this spring that line up to tell one story. Local service businesses have a new audience problem and a new audit to run this week.

The first study is BrightLocal's 2026 Local Consumer Review Survey. It runs every year on a representative panel of US adult consumers. This year the panel was 1,002 people, polled via SurveyMonkey. The headline number is the jump in how often consumers say they use ChatGPT or a similar AI tool to research a local business. In 2025 that number was 6 percent. In 2026 it is 45 percent. BrightLocal calls AI the third most popular source of local business recommendations, behind only Google Search and Google Maps.

The second study is SOCi's 2026 Local Visibility Index. SOCi looked at 350,000 brand locations across 2,751 multi-location companies. They asked the major AI assistants to recommend a business in a category and noted which locations were named. ChatGPT named 1.2 percent of brand locations. Gemini named 11 percent. Perplexity named 7.4 percent. The same set of brands appears in Google's local 3-pack 35.9 percent of the time.

Read together, the two studies make a clear point. Almost half the country is now asking ChatGPT to recommend a local business. ChatGPT, on the operator side, names roughly one in eighty locations. The gap between the two numbers is the question every local service business owner can answer for themselves this week, with a free audit they run on their own phone.

Why ranking well on Google does not cover this

The first thing to know is that ChatGPT does not run the Google local 3-pack ranking algorithm. SOCi reports the overlap between brands that perform in traditional local search and brands AI assistants recommend at roughly 45 percent. More than half of well-ranking Google brands do not show up in an AI assistant answer for the same category and area.

ChatGPT picks businesses differently. It reads what third-party websites say about a category and a location, weighs how often each business shows up across those sources, and names a small number of businesses in the answer. The websites that count are Google Business Profile, Yelp, the BBB, industry directories, online reviews, press write-ups, and mentions in forums and blogs. A business that has spent the past five years ranking well in the local 3-pack might have very little presence on any of those websites outside of Google itself.

For the operator, that means a hidden problem. The work that built the Google ranking is not the same work that puts the business in the ChatGPT answer. There is a real chance the operator who feels covered on Google has almost no presence in the answer their customer is seeing this week.

The five-minute audit to run this week

The audit is the same one your customer would run. The operator does it on their own phone or laptop.

Step one. Open ChatGPT in a private tab so the answer is not personalized to your account. Ask it the question your customer would type. "What is the best dentist in my area for an implant" if you run a dental practice. "Which HVAC company near me does heat pump installs" if you run an HVAC shop. "Where should I go for jaw Botox in my area" for a medspa. Keep the wording natural.

Step two. Read the answer. ChatGPT will usually name three to five businesses with a short description of each one. Note whether your business is in the list. If it is, read the description and note what got said. If it is not, read the description of the businesses that were named, and check the sources panel at the bottom of the response. The panel lists the URLs the assistant pulled from to build the answer.

Step three. Repeat the question on Perplexity and on Gemini. Each assistant uses a different mix of sources. Perplexity leans on news write-ups and review aggregators. Gemini leans on Google Business Profile data and is the easier place for a 3-pack incumbent to show up. The picture across all three is the picture your customer will form across the week as they cross-check the answer.

Step four. Write down what you saw. Which assistants named your business, which named your competitor, what was said in each answer, and which sources got cited. The remediation depends on the verdict. A missing business needs new presence on the third-party websites the assistant trusts. A thin description needs better content on those same websites. A wrong description needs correction at the source the assistant cited. Each variant points to a different piece of work for next week.

The audit takes five minutes per assistant. The findings open the next conversation about the off-site work that earns a place in the answer.

What the audit usually shows for a small service business

We have run this same audit against several hundred small service businesses over the past nine months. The pattern is consistent enough to call out for any operator running the audit on themselves this week.

The most common result is that ChatGPT names a directory or a national chain at position one for the category, then names two or three local competitors of the operator running the audit, then names a generic operator description without a business name in the last slot. The operator's own business is usually not in the answer. The sources that fed the answer are usually Google Business Profile, Yelp, and one industry directory.

The second most common result is that Gemini names the operator running the audit. Gemini leans on Google Business Profile more heavily than ChatGPT does, so an operator who is a 3-pack incumbent on Google is more likely to show up there. That single appearance is enough to give the operator a reason to keep the Gemini answer healthy, since most customers who cross-check will land on the assistant that already named them.

The third common result is that Perplexity names a local competitor of the operator running the audit and quotes a Reddit thread or a review aggregator in the source panel. Perplexity is the easiest assistant to read source-attribution on, and the source panel often points at the exact off-site source the operator was missing.

The three patterns together mean the work for most operators is the same. The on-site material is already in good shape, since it had to be in good shape to earn the Google 3-pack. The off-site material on Yelp, the BBB, and industry directories is usually weaker, and ChatGPT in particular relies most heavily on those pages to build its answer. The work to do this month is to look at each of those source pages, see what shows up there, and add what is missing.

How long the gap is likely to stay open

BrightLocal has been running the LCRS for fifteen years and reports that an annual change of this size in a consumer behavior is rare. The jump from 6 percent to 45 percent of consumers using ChatGPT for local business research in one year is the largest one-year change in the survey since the smartphone-driven shift in 2014.

A change that fast usually keeps moving for another year before it levels off. Which means most local service businesses have a window of a few quarters to do the work before the answer hardens around the businesses that did the off-site work first. Once an assistant has named a small set of businesses in a category and area, future answers tend to keep naming the same set, because the trust signals that fed the first answer keep feeding the next one.

The work today is small enough to fit on a Tuesday afternoon. The five-minute audit, four or five source pages to fix on Yelp and the BBB and a couple of industry directories, and a habit of running the same audit again a month from now to see what changed. The operator who runs the audit early gets the benefit of those edits across every customer query the assistant runs next week and next month.