AI Local Report, October 2026: AI changed its #1 local pick in 52% of markets in one month
We put 21,600 questions to 12 AI assistants across 600 local markets in 30 US metros. In the 245 markets we had also measured in September, the business the assistants recommended first had changed in about half of them a month later. This is the first AI Local Report from AI Syndicate Research. Every number in it was measured by us, and every number is free to cite.
Between September and October 2026, AI assistants changed their #1 local business in 52% of 245 US markets (95% range 46–58%), and 23% of September's top-10 businesses fell out of the top 25. Only 22% of markets had one #1 across three ways of asking, and more than half of all citations (54%) went to about 280 review sites, directories and other platforms.
Executive summary
Source: AI Syndicate Research, AI Local Report, October 2026 (aisyndicate.com/research)
Key findings
- AI's first pick changes about half the time. The #1 business changed between September and October in 52% of the 245 markets measured both months (95% range 46–58%). Under four stricter tests the figure runs from 46% to 61%.
- Leaders with broad support hold on. A September #1 named by ten or more of the twelve assistants was still #1 in October 74% of the time. A #1 named by five or fewer held 38% of the time.
- Even the leader is missing from most answers. In a typical market the #1 business is named in 43% of the 36 answers. Most answers name someone else.
- There is no single AI ranking. Only 22% of markets have the same #1 for "best", "recommended" and "affordable". The "affordable" #1 matches the "best" #1 in just 28% of markets.
- The assistants broadly agree with each other. A typical market's #1 is named by 8 of 12 assistants. Gemini named the #1 that the other eleven agreed on in 88% of markets; Perplexity and Google AI Overviews in about half.
- A few hundred platforms supply most of the evidence. About 280 sites, each cited in five or more markets, took 54% of all citations. Websites we could identify as a ranked business's own took 11%. Yelp is the most-cited site overall, but in 13 of 20 trades a different site leads, such as Carfax for auto repair, Zillow for real estate and EnergySage for solar.
What it means. A single AI answer about a local business is one reading, not a verdict. Position is better judged across months, across assistants and across the different ways customers ask. Businesses that kept a strong position tended to be named by many assistants and to appear on the third-party sites those assistants cite. A business's own website is only one part of that footprint. These are observed patterns, not proof of cause.
Scope. 600 markets: 30 metros in Florida, California, New York and Texas, times 20 types of business. 12 assistants, 3 fixed questions each, 21,600 questions asked between 1 and 5 October 2026. Month-to-month comparisons use only the 245 Florida and California markets measured in both editions. The method, its limits and what we hold back are set out at the end.
What does AI Syndicate Local measure?
AI Syndicate Local is a public index of which local businesses AI assistants recommend. Once a month we ask 12 assistants the same three questions about a type of business in a metro area: the best one, the one to recommend to someone who needs one, and an affordable one nearby. A market is one metro area and one type of business, so 36 answers go into each market.
A business is ranked only when it is named in at least two answers that are independent of each other: two different assistants, or one assistant answering two different questions. A name that appears in a single answer is counted but not ranked, because one answer may be a guess. Businesses are ordered by how many answers name them, then by how many assistants do. Each market's published ranking lists up to 25 businesses.
How the twelve are asked:
- Live web search: ChatGPT, Claude, Gemini and Perplexity answer with web search on. Google AI Overviews is the AI Overview Google shows for the question. Microsoft Copilot is the Copilot answer Bing shows; when Bing shows none, we reconstruct one from the same Bing results.
- Model memory only: Grok, Meta AI, DeepSeek and Mistral answer from what the model learned, because no consumer live-search access is available for them.
- Reconstructed voice answers: Siri and Alexa cannot be asked programmatically. We reconstruct their answers from the kinds of sources each draws on, in the short, single-pick style each speaks in. Read them as close approximations, not recordings.
Show the numbers
| Edition | Markets |
|---|---|
| August 2026 | 105 |
| September 2026 | 245 |
| October 2026 | 600 |
Source: AI Syndicate Research, AI Local Report, October 2026 (aisyndicate.com/research)
Of October's 21,600 questions, 20,044 (93%) returned an answer. The edition ranks about 12,700 businesses. The question wordings have not changed since the index launched in August 2026, so editions can be compared.
How often does AI change its #1 local recommendation?
About half the time. In the 245 markets measured in both months, the business the assistants named most in September was still #1 in October in 48% of markets. In the other 52%, the top spot went to someone else.
Show the numbers
| October position | Share of markets |
|---|---|
| Still #1 | 48.2% |
| Slipped to #2 or #3 | 31.4% |
| Slipped to #4–#10 | 11.8% |
| Slipped to #11–#25 | 4.5% |
| Out of the top 25 | 4.1% |
Source: AI Syndicate Research, AI Local Report, October 2026 (aisyndicate.com/research)
Four in five (79.6%) of September's #1 businesses were still in the top three, and only 4% fell out of the top 25. Further down, the list moved more. Of all businesses in a September top ten, 23% were out of the top 25 in October, and 24% of October's top-ten businesses had not been in September's top 25. A business ranked in both months typically moved 3 places.
How robust is the 52%?
Some changes at the top are close calls. In 16% of September markets the #1 was tied with #2 on answers, and in about 5% the two were tied on assistants too, so the published order fell back to the alphabet. Between the two months we also changed how we reach some assistants (see the method notes). So we re-ran the comparison four ways.
Depending on the test, the #1 changed in 46% to 61% of markets: about half. Ranking each market on only five assistants gives it fewer answers and a noisier order, which is the likeliest reason that test shows more change.
Florida has now been measured three times, which shows whether the movement settles. It has not settled yet.
Show the numbers
| Comparison | Share of markets |
|---|---|
| Same #1, August to September | 57% |
| Same #1, September to October | 50% |
| Same #1 in all three editions | 30% |
Source: AI Syndicate Research, AI Local Report, October 2026 (aisyndicate.com/research)
Of the businesses in an August Florida top ten, 74% were still in the top 25 in October.
What separates a #1 that holds from one that doesn't?
Breadth of support. September leaders that more assistants agreed on were far more likely to still be #1 in October.
The size of the lead tells the same story. A September #1 that was tied with #2 on answers held 33% of the time. One that led by six or more answers held 82% of the time. National brands held more often than independents. When September's #1 was a brand in the top three of five or more October markets, it was still #1 in 68% of cases (34 markets), against 45% for everyone else.
The ten-or-more group (19 markets, 95% range 53–95%) and the national-brand group (34 markets, 53–82%) are small, so read their exact figures with care. The direction is consistent across all three measures. These are associations. They show which leaders held, not why.
Which types of local business have the most stable AI recommendations?
Stability differs a lot by trade. The chart shows, for each type of business, the share of its compared markets that kept the same #1 from September to October. Beside it is how many of the 12 assistants named the #1 in a typical October market. Each type has only 12 or 13 compared markets, so read each row as a hint, not a verdict.
Agreement in one month does not guarantee stability the next. Hotels had the broadest agreement, with 10.3 assistants on the #1 on average, and still changed #1 in 58% of markets. Speeding ticket lawyers and roofing had below-average agreement and were the most stable.
National brands anchor some trades. In October, 14% of all top-three placings went to a brand that is in the top three in five or more markets. In the 245 compared markets that share rose from 12% in September to 15% in October. The brands most often in a top three were Renewal by Andersen, Orkin, Terminix, Servpro, PuroClean and Two Men and a Truck. In pest control and water damage restoration a national brand is #1 in about two thirds of markets; in window replacement and moving, about three in ten. Across all trades it is 16%.
The full rankings for each trade are in the index: Auto repair, business litigation lawyers, cosmetic dentists, criminal defense lawyers, electricians, garage door repair, HVAC and air conditioning, hotels, med spas, moving companies, personal injury lawyers, pest control, plumbers, real estate teams, restaurants, roofing, solar installation, speeding ticket lawyers, water damage restoration, window and door replacement.
Do ChatGPT, Gemini, Perplexity and the other assistants recommend the same businesses?
Mostly, but not fully. In a typical October market the #1 business was named by 8 of 12 assistants. In 84% of markets at least six named it. All twelve named the same #1 in only 16 of 600 markets (2.7%).
Show the numbers
| Assistants naming the #1 | Markets |
|---|---|
| 2 | 1 |
| 3 | 4 |
| 4 | 30 |
| 5 | 59 |
| 6 | 88 |
| 7 | 106 |
| 8 | 95 |
| 9 | 89 |
| 10 | 67 |
| 11 | 45 |
| 12 | 16 |
Source: AI Syndicate Research, AI Local Report, October 2026 (aisyndicate.com/research)
Lower down the rankings, support is thinner. 34% of ranked businesses were named by exactly two assistants, and 20% by a single assistant answering two different questions.
The next chart asks a stricter question of each assistant. Leave that assistant out, find the business the other eleven named most widely, and check whether the assistant named it too. This avoids counting an assistant's own vote toward the consensus it is measured against.
Gemini is the closest thing to a bellwether: it named the other eleven's choice in 88% of markets. Copilot and Claude follow. Perplexity and Google AI Overviews agreed only about half the time, which usually means they were naming businesses the others were not. A business that leads in Gemini is not necessarily leading where its customers ask.
Two cautions. First, these are shares of markets, not of users; an assistant with few users that disagrees with the rest matters less to a business than one with many. Second, the twelve are not fully independent witnesses. Several share underlying models or search indexes, and the reconstructed voice answers draw on the same kinds of sources as the others, so agreement between some pairs overstates independent confirmation.
Does the way you ask change which business AI recommends?
Yes, more than most people expect. We ask each market three ways: the best one, the one to recommend to someone who needs one, and an affordable one nearby. In October, only 22% of markets had the same business at #1 for all three wordings (133 of the 597 markets where all three were answered).
Some of that is close calls. With only twelve answers per wording, the top of a wording's list is often a tie; 28% of the time the #1 was tied with #2 on both answers and assistants. Leaving out every market with such a tie, 39% had one #1 across all three wordings. Most markets still do not.
The "affordable" question changes the list most. Its #1 matches the "best" #1 in only 28% of markets. The overall leader usually stays visible: in 82% of cases it is in a wording's top three. But a customer who asks about price is often shown a different first choice from one who asks about quality. There is no single AI ranking to win. There is one per way of asking.
Which websites do AI assistants cite when recommending a local business?
Mostly review sites, directories and other platforms. The live-search assistants returned 37,355 citations across the 600 October markets. We grouped every citation by how widely its site is cited.
| Where the citations went | Share of all citations |
|---|---|
| About 280 sites cited in five or more markets: review sites, directories, "best of" lists, forums | 54% |
| Websites we could identify as a ranked business's own | 11% |
| Sites cited in only one to four markets: local news, smaller directories, and some businesses' own sites we could not match | 20% |
| Citations beyond the 25 most-cited sites we record by name in each market | 15% |
So more than half of the evidence comes from a few hundred platforms that recur from city to city, and a business's own website is a small part of it.
The long tail is long. No single site reaches 5% of citations, and a typical market draws on 35 different source sites. What dominates depends on the trade. Yelp is the most-cited site in 7 of 20 trades; in the other 13, a site specific to the trade leads.
| Type of business | Most-cited site | Its share of the trade's citations | Next two |
|---|---|---|---|
| Auto repair | Carfax | 10.9% | Yelp, RepairPal |
| Business litigation lawyers | Super Lawyers | 8.7% | Best Lawyers, Avvo |
| Cosmetic dentists | Yelp | 5.2% | Zocdoc, Opencare |
| Criminal defense lawyers | Justia | 6.9% | Avvo, Super Lawyers |
| Electricians | Yelp | 6.0% | Thumbtack, Angi |
| Garage door repair | Yelp | 6.6% | Thumbtack, Expertise.com |
| Hotels | Tripadvisor | 11.2% | Expedia, Hotels.com |
| HVAC and air conditioning | Yelp | 5.6% | ConsumerAffairs, ServiceAgent |
| Med spas | Yelp | 6.5% | TrustAnalytica, Medical Spa Locator |
| Moving companies | MoveBuddha | 8.1% | Great Guys Moving, FreightWaves |
| Personal injury lawyers | Justia | 5.9% | Super Lawyers, Avvo |
| Pest control | Yelp | 6.0% | ConsumerAffairs, Thumbtack |
| Plumbers | Yelp | 5.8% | Angi, Thumbtack |
| Real estate teams | Zillow | 9.4% | HomeLight, FastExpert |
| Restaurants | Tripadvisor | 9.8% | OpenTable, Wanderlog |
| Roofing | Yahoo Local | 5.6% | BBB, Yelp |
| Solar installation | EnergySage | 11.9% | EcoWatch, ConsumerAffairs |
| Speeding ticket lawyers | Avvo | 10.4% | Justia, FindLaw |
| Water damage restoration | BBB | 7.6% | Expertise.com, Yelp |
| Window and door replacement | EcoWatch | 7.6% | Yelp, Expertise.com |
Shares are of all citations in each trade. Ranked businesses' own websites, and dictionary sites that assistants cite to define a word, are left out of the list. The practical reading: the sites that matter for a plumber are not the sites that matter for a solar installer or a personal injury lawyer. Reviews and Reddit matter across trades, but the trade's own directory often matters more.
What should a local business do with this?
- Judge position over several months, not one screenshot. With the #1 changing in about half of markets in a month, a single good or bad answer proves little. Three months in the same position says much more.
- Look at how many assistants name you, not only your rank. In this data, leaders named by ten or more assistants held #1 74% of the time; those named by five or fewer, 38%.
- Check every way your customers ask. Only 22% of markets have one #1 for best, recommended and affordable. Look at which question's list you are missing from.
- Look past your own website. More than half of all citations go to a few hundred review sites, directories and lists. Find the ones AI cites for your trade, using the table above as a starting point, and make sure you are present and accurate on them.
- Know whether you compete with a national brand. In pest control and water damage restoration a national brand is the first pick in about two thirds of markets, and national brands held #1 more often than independents.
These are patterns in what the assistants said, not guarantees.
How this data is used. This research feeds the AI Syndicate platform, so our clients have the leading edge. Clients see where they stand in their market's AI Local ranking, and how they have moved, inside their dashboard. Their markets are measured more often than the rest of the index, with up to three times as many answers behind each reading. What we learn across the whole index, such as which assistants name whom, which sites they cite and how each one changes, is built into the twelve-assistant measurement the platform runs for each client's own business. When an assistant changes how it answers, or how we can reach it, the platform changes with it.
How was this measured, and what can it not tell you?
Method. Each month the index asks 12 assistants three fixed questions per market, with the city named in the question and no user location or account history. A model extracts the business names from each answer, and spellings of the same business are merged. A chain's locations that share one website count as one business. Rankings follow the rule described above: at least two independent answers to be ranked, ordered by answers and then by assistants, up to 25 per market. Sources are the citations the live-search assistants return with their answers. The full method is on the methodology page.
What changed between editions. Two kinds of change sit between the September and October measurements, apart from the assistants' own answers. First, coverage: in October we re-asked any question that went unanswered the first time, which raised the share answered from 73% to 92% in the same markets. Second, access: we changed how we reach several assistants. Perplexity retired the interface we used in September, so its October answers come through its new one. Meta AI moved to Meta's own interface. DeepSeek and Mistral had fixes that let far more questions through. ChatGPT, Claude and Microsoft Copilot had smaller changes to how answers are fetched. Gemini, Google AI Overviews, Siri, Alexa and Grok were reached exactly the same way in both months, which is why one robustness test uses only those five.
Collection gaps. In this edition, Google AI Overviews and Microsoft Copilot returned usable answers in fewer than a third of California markets, against 84% to 99% of markets in the other three states. Alexa was also lower in California. We treat this as a gap in our collection, not as those assistants declining to answer. Every market is measured again each month. Agreement figures count only the markets where each assistant answered.
What this cannot tell you.
- It measures what assistants say, not what customers do. A #1 here is a share of answers, not a share of sales.
- It cannot yet separate real change in a market from the assistants' own answer-to-answer variability.
- Twelve or thirteen markets per type of business is a small sample, so trade-level figures are directional.
- The assistants were asked without a user location or history. A real customer's answer can differ.
- Siri and Alexa answers are reconstructions, and Copilot's are when Bing shows no Copilot answer.
- New York and Texas have one edition so far, so they appear only in October-wide figures.
What stays private. The public index publishes each market's rankings and sources. The month-by-month movement of individual businesses, the per-assistant answer sets, and our records of what changed on ranked businesses' websites are not published. They are the basis of our paid service and of future reports.
Statistical notes
- Unit of measurement. Every share is a share of markets, where a market is one metro area and one type of business. Each market counts equally, whatever its size.
- Questions. Each market gets three fixed questions, each asked once of each assistant, and the three count equally. Because each is asked once a month, we cannot yet separate real change from answer-to-answer variation.
- Missing answers. A question that returns no answer is left out, not counted as naming no one. An assistant's agreement figure counts only the markets where it named a ranked business.
- Ranges. The 95% ranges come from resampling markets 2,000 times. They describe uncertainty across markets, not the variability of a single answer.
- Matching businesses across months. Two entries are the same business if they share a listing, a website, or a name once "the", "and", "of" and legal suffixes such as "LLC" or "P.A." are ignored, or if one name of two or more words is the start of the other, as with Paul Davis Restoration and Paul Davis Restoration of Sarasota. Two different websites are two businesses unless the names are identical. A chain's locations that share one website count as one business.
- Typical values. "Typical" means the median market.
- Small groups. Groups of fewer than 30 markets, such as the ten-or-more-assistants group (19) and each type of business (12 or 13), have wide ranges and are directional.
How can I cite or reuse these numbers?
Quote any figure in this report with credit to AI Syndicate Research and a link to this page. You may quote up to 50 words of text, or share a single chart with its source line intact. Reproducing the report or large parts of it needs written permission. A suggested citation:
A PDF copy is available, and a plain-text Markdown version shows every chart as a table. The rankings behind the report are at aisyndicate.com/ai-local/. The next issue is out: AI Local Report, November 2026, which covers all 50 states. Journalists and researchers who need a cut of the data that is not here can write to support@aisyndicate.com.
Key figures
Every key figure in this report, in one table. The same figures are in findings.json and findings.csv, free to quote with credit. Ranges are 95% intervals from resampling markets.
| Finding | Value | 95% range | Sample |
|---|---|---|---|
| Share of markets where AI assistants' #1 local business changed from September to October 2026 | 51.8% | 46.1–57.6% | 245 markets |
| Share of markets where the #1 changed, using only the 5 assistants reached the same way in both months | 61.2% | 53.5–67.3% | 245 markets |
| Share of September's #1 businesses still in the top three in October | 79.6% | 74.7–83.7% | 245 markets |
| Share of September top-10 businesses out of the top 25 in October | 22.7% | 20.9–24.5% | 2,450 placings, 245 markets |
| Share of Florida markets with the same #1 in August, September and October 2026 | 30.5% | 20–39% | 105 markets |
| Share of September #1s named by 10 or more of 12 assistants that were still #1 in October | 73.7% | 52.6–94.7% | 19 markets |
| Share of September #1s named by 5 or fewer assistants that were still #1 in October | 38.5% | 25–51.9% | 52 markets |
| Share of answers that name the market's #1 business, typical market | 43% | 600 markets (median) | |
| Share of markets with the same #1 for the best, recommended and affordable questions | 22.3% | 18.8–25.6% | 597 markets |
| Assistants naming the #1 business in a typical market | 8 of 12 | 600 markets (median) | |
| Share of markets where Gemini named the #1 that the other eleven assistants agreed on | 87.5% | 85.6–89.6% | 599 markets |
| Share of markets where Perplexity named the #1 that the other eleven assistants agreed on | 50% | 46.9–54.1% | 586 markets |
| Share of citations going to about 280 sites cited in five or more markets | 53.5% | 52.4–54.7% | 37,355 citations, 600 markets |
| Share of citations going to websites identified as a ranked business's own | 11.4% | 10.7–12.1% | 37,355 citations, 600 markets |
Sources
- AI Syndicate Local: the October 2026 edition (600 markets), AI Syndicate. Read Oct 5, 2026.
- AI Syndicate Local methodology: how the index is measured, AI Syndicate. Read Oct 5, 2026.
Questions people ask about AI local recommendations
How often does AI change its recommendation for a local business?
In AI Syndicate Research's measurement of 245 US markets, the #1 business recommended by 12 AI assistants changed between September and October 2026 in 52% of markets (95% range 46–58%). 23% of September's top-10 businesses had fallen out of the top 25 a month later.
What keeps a business at #1 in AI recommendations?
In AI Syndicate Research's October 2026 data, leaders named by many assistants held on far more often. A September #1 named by ten or more of twelve assistants was still #1 in October 74% of the time, against 38% for one named by five or fewer. This is an association, not proof of cause.
Do ChatGPT, Gemini and Perplexity recommend the same local businesses?
Mostly. In October 2026 a typical market's #1 was named by 8 of 12 assistants. Measured against the other eleven, Gemini named the consensus #1 in 88% of markets, ChatGPT in 67% and Perplexity in 50%. All twelve agreed in only 2.7% of markets.
Which websites do AI assistants cite for local business recommendations?
In October 2026, about 280 review sites, directories and other platforms, each cited in five or more markets, took 54% of all citations. Websites identifiable as a ranked business's own took 11%. Yelp was the most-cited site overall at 4.4%, but in 13 of 20 trades a trade-specific site led, such as Carfax for auto repair, Zillow for real estate and EnergySage for solar.
Does the wording of the question change which business AI recommends?
Yes. Only 22% of markets had the same #1 business for the best one, the one to recommend, and an affordable one nearby. The affordable question's #1 matched the best question's #1 in only 28% of markets.
Can I quote these numbers?
Yes. Quote any figure with credit to AI Syndicate Research and a link to this report. The rankings behind them are at aisyndicate.com/ai-local/.
Related pages.
See who AI recommends in your market.
The index is free and public: pick your state, metro and trade to see which businesses the 12 assistants name, and which sites they cite. Or run the free check to see where your own business stands.