Written by Niladri Sarkar, CMO at Synup.com
How a local Google search is worded is closely associated with whether an AI Overview appears, and the pattern varies substantially by business category. In Synup’s new study, What’s Ranking in Local Search: 2026, an AI Overview appeared on 4% of desktop results pages for “[category] near me” searches and on 75% of pages for searches that named a specific need. The other query forms fell between those extremes.
Here’s how we got those numbers. In September 2026 my team pulled 53,900 Google results pages covering 49 kinds of local business, plumbers and dentists and accounting firms among them. Three query forms were run on desktop across the five largest cities in every US state, 250 cities in all. A fourth form, the qualified query, ran in the largest city per state, so 50 cities, with two qualifiers per category. A separate mobile run repeated the plain “[category] [city]” query only. We used browser-rendered collection, which captures features that raw-HTML scraping misses, though not everything (more on that in the method note at the end).
AI Overview frequency varies sharply by search wording
On desktop, the AI Overview rate by query form was:
- “[category] near me”: 4.0% (12,250 pages)
- “[category] [city]”: 8.6% (12,250 pages)
- “best [category] in [city]”: 27.0% (12,245 pages)
- “best [category] in [city]” plus a specific need, such as “for water heater repair”: 75.4% (4,898 pages, 50 cities)
The qualified sample is narrower, so the 75% figure rests on fewer cities than the other three. The first three query groups, which covered the same cities, also showed a clear increase, and that explains why people argue about this. A study that only ran “near me” queries would report AI Overviews as a rounding error in local search. One that only ran specific queries would report them on three pages in four. Each would be right about its own sample.
The Local Pack was common across every query form
The three-business map pack appeared on 91 to 97% of desktop pages whichever wording was used. It was the most stable feature we measured.
“Best” searches returned packs with a different rating distribution. Packs on “best” searches averaged 4.79 stars against 4.67 on the plain search, and the share of pack businesses rated 4.8 or higher went from 53% to 67%. Average review count moved only slightly, from about 233 to 251. The results suggest that rating deserves particular attention for “best” searches, but they don’t establish a threshold Google applies or show that review volume is unimportant.
Around three in ten pack listings we observed had no direct website link.
Business websites and profiles are prominent citation sources
We parsed 8,540 of the desktop AI Overviews to see what they contain and what they cite. On “best” and specific-need searches, six in ten included business cards: a name, a one-line description, a rating, a phone number, and the Call, Directions and Website buttons drawn from the business’s Google profile. Two or three per answer. (On the pages we spot-checked, the businesses in those cards weren’t always the same as the ones in the map pack below, which is a separate question we’re looking at now.)
Between 45% and 61% of citations, depending on query form, pointed at business websites, parent-brand websites or Google Business Profiles. The most frequently cited domain across the parsed AI Overviews was google.com, 13,777 times, and every one of those was a Business Profile link. Directories of all kinds accounted for 16 to 24% of citations. Reddit accounted for 4 to 6%.
That makes the profile and the website the first places to audit for accuracy and completeness, since they’re what the AI Overview is drawing on most often.
The mix of organic results stayed similar
We ran the same “near me” queries in 2024, which gives us a two-year comparison for the organic links. Business websites held 58% of those results then and hold 57% now, while directories went from 25% to 24%. So the composition of the organic list barely moved, even with an AI Overview added to the page above it. What that comparison can’t tell us is whether individual rankings, clicks or traffic stayed stable, since the study measures what appears on the page rather than what happens after.
Business category also makes a substantial difference
AI Overviews were rare for restaurant and cafe “best” searches, at 1.2% and 0.4% of pages. Rates were much higher in several other categories: home builders 80%, banks 69%, accounting firms 69%, financial advisors 62%.
Some categories with low AI Overview rates showed frequent forum blocks. Dentists had an AI Overview on 3.6% of “best” pages and a Discussions and Forums block on 62%. Vets and gyms followed a similar shape. Our data doesn’t identify which forum each block draws from, so that’s a separate observation from Reddit’s organic presence, which was on about half of “best” pages at an average organic position of about 3.
Ads ran the other way. They were most common on “near me” searches in professional categories, with marketing agencies at 89% of pages and lawyers at 78%, and close to absent for restaurants and cafes.
What to do with this if you run a business
The study measures what appears on the page. It doesn’t measure why, and being cited, appearing in a pack and winning the customer’s click are three different things. With that said, a few actions follow from the findings:
- Search for your own category the four ways above and see which surfaces show up. The answer differs by category more than most people expect.
- Audit your Google Business Profile and website for accurate service and location information, since those were the most frequent AI Overview citations.
- Check your category’s AI Overview rate in the study’s explorer before deciding how much attention it deserves. For a cafe and a home builder, the answer is very different.
A note on method
Collection method materially affects reported AI Overview rates. In our collection tests, raw-HTML scraping detected substantially fewer AI Overviews than browser-rendered collection, so comparisons between studies should account for rendering method, query mix and category coverage. Our own method has gaps too: Local Services Ads and sponsored pack slots didn’t appear in any of 41,650 desktop pages, but live checks in September 2026 found LSAs on two of three high-value service searches and a sponsored pack entry on all three. Those are reported as limitations, not as findings.
Explore your business category and download the supporting data in Synup’s full local search study.
Niladri Sarkar is CMO at Synup, a local marketing platform used by agencies and multi-location brands to manage listings, reviews and search visibility.
