# How close is the race to be AI's local pick? what it takes to rank in 1,400 US markets

> When someone asks an AI assistant for a plumber, a lawyer or a restaurant, a handful of names come back. How hard is it to be one of them? We read AI's rankings in 1,400 US local markets, built from 43,587 answers by 12 AI assistants, and measured what it takes to rank: how many answers name the business at #1, #3 and #10, how far apart the top businesses are, how many assistants have to agree, and which markets are still wide open. Every number was measured by us and is free to cite.

Updated October 14, 2026 · AI Syndicate team · https://www.aisyndicate.com/research/how-close-is-the-ai-local-race-october-2026/

**The findings in one paragraph:** In AI Syndicate Research's October 2026 study of 1,400 US local markets and 43,587 answers from 12 AI assistants, the business at **#3** was named in a median of **8** of about 31 answers, and #1 in 14. In **50%** of markets #1 led #2 by two answers or fewer, and **#3 and #4 were one answer apart or less in 64%**. **76%** of top-three places went to businesses five or more assistants named. **41%** of markets were wide open; 20% had a clear leader.

**Cite this:** AI Syndicate Research, "How Close Is the Race to Be AI's Local Pick? What It Takes to Rank in 1,400 US Markets," October 2026, measured 1–12 October 2026, https://www.aisyndicate.com/research/how-close-is-the-ai-local-race-october-2026/

**Use and reuse:** © 2026 AI Syndicate. You may quote up to 50 words, or share a single chart with its source line, with credit and a link to this page. Reproducing the report or large parts of it needs written permission from support@aisyndicate.com. The rankings behind it are open data under a CC BY 4.0 licence.

**Download the PDF:** https://www.aisyndicate.com/research/how-close-is-the-ai-local-race-october-2026/how-close-is-the-ai-local-race-october-2026.pdf

## Summary

**Figure 1. What it takes to be one of AI's local picks, in six numbers**

| Item | Value |
| --- | --- |
| Different businesses the 12 assistants named in a typical market | 51 |
| Of the businesses named in a market came up in just one answer | 62% |
| Answers, out of about 31, that put a business at #3 | 8 |
| Of markets where #1 led #2 by two answers or fewer | 50% |
| Of markets where #3 and #4 were one answer apart or less | 64% |
| Of top-three spots went to a business five or more assistants named | 76% |

_Source: AI Syndicate Research, How Close Is the Race to Be AI's Local Pick?, October 2026 (aisyndicate.com)_

### Key findings

1. **AI names a crowd, then agrees on a few.** In a typical market the 12 assistants named **51** different businesses across about 31 answers. **62%** of them came up in only one answer. Twenty made the ranking.
2. **The bar to rank is low and the steps are small.** The median #1 was named in **14** answers, #3 in **8** and #10 in **3**. From #10 to #3 is a median of five more answers.
3. **The top is close.** In **50%** of the 1,400 markets, #1 led #2 by two answers or fewer, and the two were tied in **13%**. #1 led by five or more in 29%.
4. **The edge of the top three is closer still.** #3 and #4 were tied in 31% of markets and one answer apart or less in **64%**. In the median market, two businesses below #3 were within two answers of it.
5. **Breadth across assistants is what ranks.** The median #1 was named by **7** assistants, #3 by 5 and #10 by 2. **76%** of top-three spots went to a business five or more assistants named and **1%** to one named by two or fewer. A business one assistant names on every question can rank, but almost never above #10.
6. **Two in five markets are wide open.** We call a market wide open when its #1 was named in under a third of answers or led by one answer or less: **41%** of markets. **20%** had a clear leader, named in half the answers and five or more ahead.
7. **It depends on the trade.** **69%** of criminal defense markets were wide open, and 54% of plumber, med spa and real estate markets. Garage door repair had a clear leader in **50%** of markets and window replacement in 43%.
8. **Clear leaders are mostly local.** A national brand was #1 in 13% of all markets and 15% of clear-leader markets. Most clear leaders are local businesses.
9. **The assistants that search the web name the #1 more often.** Gemini named the market's #1 in **93%** of the markets where it named a ranked business, Microsoft Copilot in 88% and Claude in 87%. The four that answer from memory did so in 58% to 67%.
10. **"Affordable" is a different race.** The overall #1 was also #1 for "who are the best" in 69% of markets and for "I need a … who do you recommend" in 66%, but for "best affordable" in only **44%**. One business topped all three questions in 23%.

## What did we measure, and how?

Each month AI Syndicate Local asks 12 AI assistants three questions about a type of business in a metro area: who are the best, who do you recommend, and a good affordable one nearby. For a plumber it reads like "I need a plumber in Houston, TX. Who do you recommend?" The October 2026 edition asked about **20 kinds of local business** in **70 metro areas** across 14 states (Alabama, Alaska, Arizona, Arkansas, California, Colorado, Connecticut, Delaware, Florida, Georgia, Hawaii, Idaho, New York and Texas): **1,400 markets**, measured 1–12 October 2026.

43,587 answers came back, a median of 31 per market. Counting each business once in each market, they named 75,235 businesses. This paper does not ask *why* AI picks a business. It asks **how many times AI has to name a business for it to rank, and how far apart the businesses at the top are.**

| Term | What it means here |
| --- | --- |
| Answer | One assistant's reply to one of the three questions in one market |
| Named in an answer | The business appears by name in that reply. Each answer counts once per business |
| Ranked | Named in at least two answers in the market. Ranked businesses are ordered by how many answers named them |
| Top three | The three most-named businesses in a market |
| Lead | How many more answers named #1 than named #2 |

The twelve fall into three groups by how they answer:

- **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:** Grok, Meta AI, DeepSeek and Mistral answer from what the model learned, because no consumer live-search access is available for them.
- **Voice:** Siri and Alexa cannot be asked programmatically, so we reconstruct their answers in the short, spoken style each uses.

In 202 markets (99 in Arizona, 37 in Arkansas, 32 in Alaska, 31 in Alabama and 3 in Georgia) the edition asked nine assistants, without Google AI Overviews, Microsoft Copilot and Alexa. Every headline figure is within about a point of the same figure for the 1,198 markets where all 12 were asked.

## How many businesses does AI name in one market?

A lot of them, and most only once. In the median market the 12 assistants named **51** different businesses (the middle 80% of markets ran from 34 to 76). **20** were named in two or more answers and made the ranking.

In the 1,056 markets where fewer than 25 businesses ranked, so that nothing was cut by the 25-place cap, **62%** of the businesses named came up in exactly one answer. One assistant, on one question, named them, and none of the other answers did.

**Figure 2. Different businesses AI named in a market, by trade**

_Median per market across about 31 answers from 12 AI assistants. 70 markets per trade, October 2026._

| Trade | Businesses named |
| --- | --- |
| Restaurants | 90 |
| Auto repair | 79 |
| Real estate agents | 64 |
| Med spas | 62 |
| Roofing | 58 |
| HVAC | 58 |
| Moving companies | 55 |
| Plumbers | 55 |
| Electricians | 54 |
| Cosmetic dentists | 51 |
| Hotels | 50 |
| Criminal defense lawyers | 49 |
| Pest control | 49 |
| Window replacement | 45 |
| Water damage restoration | 45 |
| Solar installation | 44 |
| Business litigation lawyers | 43 |
| Personal injury lawyers | 41 |
| Speeding ticket lawyers | 40 |
| Garage door repair | 37 |

_Source: AI Syndicate Research, How Close Is the Race to Be AI's Local Pick?, October 2026 (aisyndicate.com)_

Restaurants (90) and auto repair (79) had the biggest crowds, garage door repair (37) and speeding ticket lawyers (40) the smallest. A crowd of one-off names is not a ranking. It is the reason a single AI answer is a poor guide to where a business stands: most of the names in it will not be in the next one.

## How many answers does it take to rank?

Fewer than most people expect. In the median market, about 31 answers came back. The business at **#1** was named in **14** of them (44%). The business at **#3** was named in **8** (26%). **#10** was named in **3** (10%). The last ranked place takes two.

**Figure 3. How many answers named the business at each place**

_Median across 1,400 markets, October 2026. A median of 31 answers came back per market._

| Place in the market | Answers that named it |
| --- | --- |
| #1 | 14 |
| #2 | 10 |
| #3 | 8 |
| #5 | 6 |
| #10 | 3 |
| Last ranked | 2 |

_Source: AI Syndicate Research, How Close Is the Race to Be AI's Local Pick?, October 2026 (aisyndicate.com)_

The middle half of markets needed 11 to 17 answers for #1, 6 to 10 for #3 and 3 to 4 for #10. From #10 to #3 is a median of **five more answers**. With 12 assistants and three questions each, that is a handful more assistant-question pairs saying a business's name.

## How close is the race at the top?

Very close. In **12.5%** of markets #1 and #2 were named in the same number of answers. In another **20%** #1 led by one answer, and in 17% by two. Taken together, in **50%** of markets one or two answers separate the top two. #1 led by five or more answers in **29%**, and by ten or more in 6%.

**Figure 4. How many more answers named #1 than named #2**

_Share of 1,400 markets, October 2026._

| Lead of #1 over #2 | Share of markets |
| --- | --- |
| Tied | 12.5% |
| One answer | 20.2% |
| Two | 17.1% |
| Three or four | 21.6% |
| Five to nine | 22.1% |
| Ten or more | 6.4% |

_Source: AI Syndicate Research, How Close Is the Race to Be AI's Local Pick?, October 2026 (aisyndicate.com)_

The edge of the top three is closer still. #3 and #4 were tied in **31%** of markets and one answer apart or less in **64%**. In the median market, two businesses below #3 were within two answers of it. Down at #10 and #11, 73% of markets had a tie.

**How a tie is broken.** When two businesses were named in the same number of answers, the one more assistants named ranks higher. In 175 markets #1 and #2 were tied on answers, and in 69 of them (4.9% of all markets) they were also tied on assistants, so the order between them came down to the alphabet. We left those markets in. They are part of why this paper speaks of a race rather than a winner.

## Does it take one assistant or many?

Many. The number of different assistants that named a business tracks its place closely (rank correlation 0.85 between answers and assistants). The median #1 was named by **7** of the assistants that answered, #3 by **5**, #5 by 4 and #10 by 2.

**Figure 5. Top-three places, by how many assistants named the business**

_4,196 top-three places in 1,400 markets, October 2026._

| Assistants that named the business | Share of top-three places |
| --- | --- |
| Five or more | 76.4% |
| Three or four | 22.5% |
| One or two | 1.1% |

_Source: AI Syndicate Research, How Close Is the Race to Be AI's Local Pick?, October 2026 (aisyndicate.com)_

**76%** of top-three places went to a business that five or more assistants named, and **1.1%** (47 of 4,196) to one that two or fewer named. The same holds for #1: 92% of #1s were named by five or more.

One assistant can get a business onto the list. **21.5%** of all ranked businesses were named by one assistant only, on two or three of its questions. But those businesses sat at #11 or below 91% of the time, and only three of them in 1,400 markets reached the top three. Each assistant tends to say a top business's name on one or two of its three questions (a median of two answers per assistant for #1, 1.5 for #10), so being named by one more assistant moves a business further than being named on one more question.

## Which assistants name the business at the top?

The assistants that search the web, mostly. For each assistant we took the markets where it named at least one ranked business and counted how often one of the names was the market's #1.

**Figure 6. How often each assistant named the market's #1**

_Share of the markets where the assistant named at least one ranked business. October 2026. Siri, Alexa and Microsoft Copilot are reconstructions._

| Assistant | Named the #1 | Markets |
| --- | --- | --- |
| Gemini | 93.3% | 1,395 |
| Microsoft Copilot | 88.4% | 517 |
| Claude | 86.7% | 1,379 |
| ChatGPT | 79.7% | 1,397 |
| Siri | 77.9% | 1,392 |
| Meta AI | 66.7% | 1,346 |
| DeepSeek | 64.1% | 1,241 |
| Alexa | 62.0% | 1,082 |
| Perplexity | 61.8% | 1,376 |
| Google AI Overviews | 61.3% | 911 |
| Grok | 60.9% | 860 |
| Mistral | 58.2% | 1,146 |

_Source: AI Syndicate Research, How Close Is the Race to Be AI's Local Pick?, October 2026 (aisyndicate.com)_

Gemini named the #1 in **93%** of its markets, Microsoft Copilot in 88% and Claude in 87%. ChatGPT (80%) and Siri (78%) followed. The assistants that answer from memory, Meta AI (67%), DeepSeek (64%), Grok (61%) and Mistral (58%), named it least, alongside Alexa (62%), Perplexity (62%) and Google AI Overviews (61%).

Two things shape this chart. The #1 is defined by all the assistants together, so an assistant that names many businesses (Claude named a median of 9 ranked businesses per market, Google AI Overviews 3) has more chances to include it. And Microsoft Copilot, Google AI Overviews and Grok named a ranked business in fewer markets (517, 911 and 860), so their figures rest on fewer markets.

**Where each assistant's other names land.** Siri's short answers name few businesses, and 52% of the ranked ones it named were in the top three; only 11% sat at #11 or below. At the other end, **43%** of Mistral's ranked names, 36% of Meta AI's, 35% of Claude's and 35% of DeepSeek's were at #11 or below: names no other assistant backed up. ChatGPT (24%), Google AI Overviews (23%) and Alexa (25%) had the fewest outside the top ten after Siri.

## Does the #1 hold across the three questions?

Not always. The three questions are separate races. In the 1,390 markets where all three produced a ranking, the market's overall #1 was also #1 for "who are the best …" in **69%**, and for "I need a … who do you recommend?" in **66%**. For "best affordable …" it was #1 in only **44%**, and in the top three for that question in 68%.

**Figure 7. How often the overall #1 was also #1 for each question**

_1,390 markets where all three questions produced a ranking, October 2026._

| Question | Also #1 | In its top 3 |
| --- | --- | --- |
| Who are the best … | 69.3% | 93% |
| I need a … who do you recommend? | 66.3% | 92% |
| Best affordable … | 44.0% | 68% |

_Source: AI Syndicate Research, How Close Is the Race to Be AI's Local Pick?, October 2026 (aisyndicate.com)_

One business was #1 for all three questions in **23%** of markets. "Affordable" brings in a different set of names, so a business that is not the overall #1 can still be the name AI gives to people asking for value.

## Which markets are wide open, and which have a clear leader?

We sorted every market into three kinds by its #1:

- **Wide open:** the #1 was named in under a third of answers, or led #2 by one answer or less. **41%** of markets.
- **Clear leader:** the #1 was named in at least half the answers and led #2 by five or more. **20%** of markets.
- **In between:** the other **39%**.

The cut-offs are ours, chosen to be easy to read. Under nearby cut-offs, criminal defense stays the most open trade and garage door repair and window replacement the least.

**Figure 8. Share of markets that were wide open, by trade**

_70 markets per trade, October 2026. Wide open: the #1 was named in under a third of answers, or led #2 by one answer or less._

| Trade | Wide open | Clear leader |
| --- | --- | --- |
| Criminal defense lawyers | 69% | 3% |
| Real estate agents | 54% | 13% |
| Med spas | 54% | 13% |
| Plumbers | 54% | 10% |
| Cosmetic dentists | 51% | 7% |
| Roofing | 49% | 7% |
| Auto repair | 49% | 7% |
| HVAC | 46% | 10% |
| Speeding ticket lawyers | 44% | 19% |
| Pest control | 43% | 14% |
| Electricians | 40% | 10% |
| Personal injury lawyers | 39% | 23% |
| Business litigation lawyers | 39% | 11% |
| Moving companies | 31% | 40% |
| Solar installation | 31% | 37% |
| Hotels | 30% | 21% |
| Restaurants | 29% | 31% |
| Garage door repair | 27% | 50% |
| Water damage restoration | 20% | 34% |
| Window replacement | 17% | 43% |

_Source: AI Syndicate Research, How Close Is the Race to Be AI's Local Pick?, October 2026 (aisyndicate.com)_

**Criminal defense lawyers** were the most open trade: **69%** of markets wide open and a clear leader in 3%. Plumbers, med spas and real estate agents were wide open in 54% of markets, cosmetic dentists in 51%. At the other end, **garage door repair** had a clear leader in **50%** of markets, window replacement in 43% and moving companies in 40%.

**Are the clear leaders chains?** Mostly not. We counted a business as a national brand when it was in the top three in five or more of the 1,400 markets: 32 names, such as Orkin, Terminix, Renewal by Andersen, Two Men and a Truck, Roto-Rooter and Servpro. A national brand was #1 in **13%** of all markets, **15%** of clear-leader markets and 13% of wide-open ones. Brands did own some trades: they were #1 in 61% of pest control markets and 49% of moving markets. But in garage door repair, where half the markets had a clear leader, a brand was #1 in 10%. Most clear leaders are local businesses that many assistants agree on.

## Does it change from state to state?

The pattern holds in every state, with some spread. California had the most wide-open markets (49%) and Alaska and Delaware the fewest (20% and 22%).

| State | Markets | Wide open | Clear leader | #1 named in (median) |
| --- | --- | --- | --- | --- |
| Alabama | 80 | 42% | 26% | 51% of answers |
| Alaska | 40 | 20% | 28% | 46% of answers |
| Arizona | 100 | 47% | 15% | 44% of answers |
| Arkansas | 80 | 41% | 22% | 45% of answers |
| California | 180 | 49% | 16% | 41% of answers |
| Colorado | 100 | 44% | 18% | 44% of answers |
| Connecticut | 80 | 36% | 10% | 44% of answers |
| Delaware | 60 | 22% | 20% | 44% of answers |
| Florida | 140 | 45% | 24% | 42% of answers |
| Georgia | 100 | 47% | 24% | 44% of answers |
| Hawaii | 80 | 32% | 30% | 49% of answers |
| Idaho | 80 | 35% | 21% | 45% of answers |
| New York | 140 | 39% | 16% | 46% of answers |
| Texas | 140 | 40% | 21% | 43% of answers |

The states with fewer markets per metro, and the markets where nine assistants were asked rather than 12, have fewer answers per market. A smaller pool of answers makes shares jump more, so read the state differences as rough.

## What does this mean for a local business?

These findings describe where the line is, not how to cross it: this paper did not test what moves a business up. Read with that in mind, they suggest five things.

1. **Look up your own market.** The [AI Local index](/ai-local/) shows the ranked businesses in each market, how many answers named each one and which assistants did. Find your place and count the gap to the place above.
2. **Count in answers, not percentages.** In the median market, #3 is eight answers and #10 is three. A business at #8 is usually a few answers from the top three, and a business that is not ranked at all is often one more answer from being on the list.
3. **Aim for more assistants, not more mentions from one.** Top-three places went to businesses five or more assistants named. The web-search assistants read the open web, so being described on the sites they read matters to them; the memory assistants answer from what the model learned, which changes slowly. One assistant naming a business on every question rarely lifts it above #10.
4. **Treat "affordable" as its own race.** The overall #1 is the affordable #1 in fewer than half of markets. A business that competes on price can be the name AI gives for value even when it is not the overall #1.
5. **Check monthly, and do not read one answer.** With two answers between #1 and #2 in half of markets, places move. 62% of the names in any single answer never show up again. Judge where you stand from many answers across all 12 assistants.

## How was this measured, and what can it not tell you?

**Method.** The October 2026 edition of AI Syndicate Local asked 12 assistants three fixed questions about each of 20 kinds of local business in each of 70 metro areas, with the city named in the question and no user location or account history. A model extracts the business names from each answer. Names are matched within a market, so the same business named two ways counts once. A business is ranked when it is named in at least two answers, and ranked businesses are ordered by the number of answers that named them, then by the number of assistants, then by name. The ranking keeps the top 25. This paper reads the published rankings for all 1,400 markets.

**How the twelve are reached.** ChatGPT, Claude, Gemini and Perplexity answer with web search on. Google AI Overviews is the AI Overview Google shows. Microsoft Copilot is the Copilot answer Bing shows, reconstructed from the same Bing results when Bing shows none. Siri and Alexa are reconstructions in the short spoken style each uses. Grok, Meta AI, DeepSeek and Mistral answer from model memory, because we cannot reach their live search; their consumer apps may answer differently.

**What this cannot tell you.**

- **Why a business ranks.** This paper measures how often and how widely a business is named, not what caused it.
- **Whether a place will hold.** Each question is asked once a month. A one-answer lead is within the range a second asking could change; we did not ask twice in the same week.
- **How a single real customer is answered.** Real people ask in their own words, from their own location and account. The three fixed questions are a sample, not every question people ask.
- **Exact figures for every assistant.** Siri, Alexa and Microsoft Copilot are reconstructions. Google AI Overviews, Microsoft Copilot and Alexa were not asked in 202 markets, and returned fewer usable answers in others.
- **Every name.** The extractor misses some names and the matcher can split one business into two spellings. Both make the crowd of one-off names look a little bigger than it is.
- **Chains, exactly.** "National brand" is a rule (top three in five or more markets), not a list we checked by hand. Chains that are named by location, such as hotel chains, do not meet it.

## Statistical notes

- **Unit.** A market is one kind of business in one metro area. Figures are medians or shares across the 1,400 markets unless the text says otherwise.
- **Ranges.** 95% ranges come from resampling the 1,400 markets 1,000 times. For example: markets where #1 led by two answers or fewer, 49.9% (47.4–52.4%); tied, 12.5% (10.7–14.3%); #3 and #4 one answer apart or less, 64.1% (61.6–66.9%); wide-open markets, 40.8% (38.3–43.4%); clear-leader markets, 20.2% (18.2–22.4%); names that came up in one answer, 61.9% (61.4–62.4%); top-three places from five or more assistants, 76.4% (74.8–78.0%); a national brand at #1, 13.3% (11.5–15.0%).
- **Robustness.** On the 1,198 markets where all 12 assistants were asked, the lead of two answers or fewer is 50.0%, wide open 40.9%, clear leader 20.1%, and the median answers for #1, #3 and #10 are 14, 8 and 3.
- **One-off names.** Measured in the 1,056 markets with fewer than 25 ranked businesses, where the 25-place cap removed nothing, so every unranked name was named exactly once.
- **Missing answers.** A question that returned no answer is left out. Shares of answers use the answers that came back.
- **Per-assistant figures.** Counted over the markets where that assistant named at least one ranked business.

## How to cite this report

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:

> AI Syndicate Research, "How Close Is the Race to Be AI's Local Pick? What It Takes to Rank in 1,400 US Markets," October 2026, measured 1–12 October 2026, https://www.aisyndicate.com/research/how-close-is-the-ai-local-race-october-2026/

A [PDF copy](/research/how-close-is-the-ai-local-race-october-2026/how-close-is-the-ai-local-race-october-2026.pdf) and a plain-text [Markdown version](/research/how-close-is-the-ai-local-race-october-2026/index.md) are available. The rankings behind the report are published under a CC BY 4.0 licence at [aisyndicate.com/ai-local/](/ai-local/), with machine-readable copies at [/ai-local/data.json](/ai-local/data.json) and [/ai-local/all.md](/ai-local/all.md). Journalists, businesses 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 one table. Data: https://www.aisyndicate.com/research/how-close-is-the-ai-local-race-october-2026/findings.json · https://www.aisyndicate.com/research/how-close-is-the-ai-local-race-october-2026/findings.csv (CC BY 4.0). Ranges are 95% intervals from resampling markets.

| Finding | Value | 95% range | Sample |
| --- | --- | --- | --- |
| Median number of different businesses 12 AI assistants named in one local market | 51 | 50–52 | 1,400 markets |
| Median number of businesses named often enough to rank (two or more answers) in one market | 20 |  | 1,400 markets |
| Share of the businesses AI named in a market that came up in only one answer | 61.9% | 61.4–62.4% | 1,056 markets |
| Median number of answers that named the #1 business in a market | 14 |  | 1,400 markets, median 31 answers each |
| Median number of answers that named the #3 business in a market | 8 | 8–8 | 1,398 markets |
| Median number of answers that named the #10 business in a market | 3 |  | 1,377 markets |
| Share of markets where #1 was named in two or fewer more answers than #2 | 49.9% | 47.4–52.4% | 1,400 markets |
| Share of markets where #1 and #2 were named in the same number of answers | 12.5% | 10.7–14.3% | 1,400 markets |
| Share of markets where #3 and #4 were one answer apart or less | 64.1% | 61.6–66.9% | 1,397 markets |
| Median number of AI assistants that named the #1 business in a market | 7 |  | 1,400 markets |
| Share of top-three places that went to a business five or more assistants named | 76.4% | 74.8–78% | 4,196 places |
| Share of top-three places that went to a business two or fewer assistants named | 1.1% | 0.7–1.5% | 4,196 places |
| Share of ranked businesses named by only one AI assistant | 21.5% | 20.8–22.2% | 27,845 ranked businesses |
| Share of markets where Gemini named the market's #1 (highest of 12 assistants) | 93.3% |  | 1,395 markets |
| Share of markets where Mistral named the market's #1 (lowest of 12 assistants) | 58.2% |  | 1,146 markets |
| Share of markets where the overall #1 was also #1 for the "best affordable" question | 44% |  | 1,390 markets |
| Share of markets that were wide open (the #1 named in under a third of answers, or led by one answer or less) | 40.8% | 38.3–43.4% | 1,400 markets |
| Share of markets with a clear leader (named in half the answers and five or more ahead) | 20.2% | 18.2–22.4% | 1,400 markets |
| Share of criminal defense lawyer markets that were wide open (most open trade) | 69% |  | 70 markets |
| Share of markets where the #1 was a national brand (top three in five or more markets) | 13.3% | 11.5–15% | 1,400 markets |

## FAQ

### How many times does AI need to name my business for it to rank in the top three?

In AI Syndicate Research's October 2026 study of 1,400 US local markets, the business at #3 was named in a median of 8 of about 31 answers from 12 AI assistants. The #1 was named in 14, and #10 in 3. Being named by more assistants mattered more than being named more often by one.

### How close is the race to be AI's #1 local recommendation?

Close. In half of the 1,400 markets studied, the #1 business was named in two or fewer more answers than #2, and the two were tied in 13%. #3 and #4 were one answer apart or less in 64% of markets.

### Can one AI assistant put a business at the top?

Rarely. 22% of ranked businesses were named by only one assistant, but 91% of those sat at #11 or below, and only three reached the top three in 1,400 markets. 76% of top-three places went to businesses five or more assistants named.

### Does ChatGPT name the same #1 as the other AI assistants?

Usually. ChatGPT named the market's #1 in 80% of the markets where it named a ranked business. Gemini did in 93%, Microsoft Copilot in 88% and Claude in 87%. The assistants that answer from memory (Meta AI, DeepSeek, Grok and Mistral) did in 58% to 67%.

### Which local trades are the most open in AI recommendations?

Criminal defense lawyers: 69% of markets were wide open, with no business named in a third of answers or the top two within one answer. Plumbers, med spas and real estate agents were wide open in 54%. Garage door repair had a clear leader in 50% of markets.

### How was this research done?

AI Syndicate Local asked 12 AI assistants three fixed questions about 20 kinds of local business in 70 US metro areas in October 2026: 1,400 markets and 43,587 answers. This paper reads the published rankings for every market. Siri, Alexa and Microsoft Copilot are reconstructions. Method and limits are on the page.

## Sources

1. [AI Syndicate Local: the October 2026 edition](https://www.aisyndicate.com/ai-local/) — AI Syndicate. Read Oct 14, 2026.
2. [AI Syndicate Local methodology: how the index is measured](https://www.aisyndicate.com/ai-local/methodology/) — AI Syndicate. Read Oct 14, 2026.
3. [AI Syndicate Local, machine-readable rankings with the assistants that named each business (CC BY 4.0)](https://www.aisyndicate.com/ai-local/data.json) — AI Syndicate. Read Oct 14, 2026.