AI Syndicate Research: measured data on how AI recommends businesses
AI Syndicate Research publishes what we measure about how AI assistants recommend businesses. Its main series is the AI Local Report, written from each monthly edition of AI Syndicate Local, our public index of which local businesses 12 AI assistants name across US markets.
- ChatGPT
- Claude
- Gemini
- Perplexity
- Google AI Overviews
- Microsoft Copilot
- Grok
- Meta AI
- DeepSeek
- Mistral
- Siri
- Alexa
- AI assistants asked
- 12
- US local markets
- 4,100
- AI answers in November
- 146,241
- To cite, with credit
- Free
Read the research.
Does AI recommend local businesses differently in Canada and the UK?
The assistants answer alike in all three countries; the sites they read do not. Yelp is cited in 92% of US markets and 22% of UK ones, where Checkatrade leads.
Local AI in 2027: nine predictions from our data
AI's #1 local pick changed in 45% of markets in one month, and every #1 was named by an assistant that searches the web. What four editions show, and nine predictions we will be marked against.
Can you trust the phone number AI gives you?
Live-search assistants' phone numbers for local businesses matched the business's own website 87% of the time. Memory-only assistants', 23%, and they give out half of all the numbers.
Do AI assistants favor national chains?
AI's #1 local pick is a national chain in 15% of 4,100 US markets. Assistants that answer from memory lead with a chain 36% of the time; the ones that search, 14%.
AI Local Report, November 2026
12 AI assistants, 4,100 US markets in all 50 states. AI's #1 local pick changed in 40% of markets in a month; a clear lead held 85% of the time.
How close is the race to be AI's local pick?
What it takes to rank in AI's local recommendations across 1,400 US markets: #3 is a median of 8 answers, #1 leads #2 by two answers or fewer in half of markets, and 41% of markets are wide open.
How real estate agents can stay ahead of AI
What 12 AI assistants said about 7,832 real estate picks: brands are named twice as often as agents, 73% of names come from one assistant, and "affordable" questions go to discount services.
How lawyers can stay ahead of AI
What 12 AI assistants said about 20,276 lawyer picks: 73% of firms are named by only one assistant, the firms they agree on have peer ratings and results, and Avvo and the state bar fill the gaps.
Why AI recommends a local business
What 12 AI assistants said about 130,996 local picks: reviews and price lead the reasons, 90% of the star ratings quoted are 4.5 or higher, and some assistants won't name a lawyer.
The 12 AI assistants compared
Two camps, different favorites: assistants that answer from memory name national brands 2.4 times as often as assistants that search the web.
AI Local Report, October 2026
12 AI assistants, 600 US markets, 21,600 questions. AI's #1 local pick changed in about half of markets in one month.
AI Syndicate Research is the publishing arm of AI Syndicate's measurement work. The AI Local Report comes out after each monthly edition of the AI Local index, compares it with earlier editions, and states its method and limits. Every figure is measured by us and free to cite with credit.
Which reports have been published?
| Report | Published | Headline finding |
|---|---|---|
| Does AI recommend local businesses differently in Canada and the UK? | 18 November 2026 | AI cited Yelp in 92% of US local markets but 22% of UK ones, where Checkatrade led (47%); US-only directories such as Angi and Avvo were almost never cited abroad |
| Local AI in 2027: nine predictions from our data | 17 November 2026 | AI's #1 local pick changed in 45% of 1,400 US markets in one month; nine predictions for November 2027, each with a number to be marked against |
| Can you trust the phone number AI gives you? | 16 November 2026 | Phone numbers from memory-only AI assistants matched the business's own website 23% of the time; from live-search assistants, 87% |
| Do AI assistants favor national chains? | 12 November 2026 | AI's #1 local pick is a national chain in 15% of 4,100 US markets; memory-only assistants lead with a chain 36% of the time |
| AI Local Report, November 2026 | 11 November 2026 | AI's #1 local pick changed in 40% of 1,400 US markets in one month; a #1 with a clear lead held 85% of the time |
| How close is the race to be AI's local pick? | 14 October 2026 | AI's #1 local pick led #2 by two answers or fewer in 50% of 1,400 US markets |
| How real estate agents can stay ahead of AI | 11 October 2026 | 46% of AI's picks for an "affordable" real estate agent were discount or online services |
| How lawyers can stay ahead of AI | 9 October 2026 | 73% of the law firms AI named in a market were named by only one of 12 AI assistants |
| Why AI recommends a local business | 8 October 2026 | 90% of the star ratings 12 AI assistants quoted for the local businesses they recommended were 4.5 or higher |
| The 12 AI assistants compared | 6 October 2026 | AI assistants that answer from memory name national brands 2.4 times as often as assistants that search the web |
| AI Local Report, October 2026 | 5 October 2026 | AI's #1 local pick changed in 52% of 245 US markets in one month |
Does AI recommend local businesses differently in Canada and the UK? is a standalone white paper on the first markets outside the US: 216 markets in 10 Canadian and 12 UK cities from the November 2026 edition, compared with the 54 largest US cities re-weighted to the same trades. It shows that the 12 assistants answer alike in all three countries but read different websites, and which review sites matter in each. It also has a PDF copy.
Local AI in 2027 is a standalone white paper drawing on all four editions from August to November 2026: it sets out what the record shows about how often AI's #1 changes, which assistants decide it and how close markets are, and makes nine predictions for November 2027 that we will mark hit or miss. It also has a PDF copy.
Can you trust the phone number AI gives you? is a standalone paper from the November 2026 edition: it checks the phone numbers and web addresses all 12 assistants gave for businesses in 4,100 US markets against the businesses' own websites and public DNS, and shows which assistants get them right. It also has a PDF copy.
Do AI assistants favor national chains? is a standalone paper from the November 2026 edition: it classes every business AI ranked in 4,100 US markets as a national chain or a local business, and shows which of the 12 assistants lead with chains, in which trades, and how local businesses win where chains usually do. It also has a PDF copy.
The November 2026 report is the first from an edition covering all 50 states: 4,100 markets, 12 assistants and 147,600 questions. It compares October and November in the 1,400 markets measured both months, follows 245 markets across three editions, and shows where national chains are AI's first pick. A PDF copy is available.
The October 2026 report covers 600 markets in Florida, California, New York and Texas, 12 assistants and 21,600 questions. It compares the September and October editions and, for Florida, three editions back to August. A PDF copy is available.
The 12 AI assistants compared is a standalone paper from the same October data: it sets the twelve assistants side by side and shows how differently each one picks local businesses. It also has a PDF copy.
Why AI recommends a local business is a standalone paper that reads what all 12 assistants wrote about 130,996 businesses they recommended in the 1,000 markets of the October edition: the reasons they gave, the ratings and prices they quoted, and when they named no one. It also has a PDF copy.
How lawyers can stay ahead of AI is a standalone paper that reads 6,227 answers about lawyers from all 12 assistants in the 200 legal markets of the October edition: whether they name a lawyer, how much they agree, the reasons and directories behind their picks, and a plan for law firms. It also has a PDF copy.
How real estate agents can stay ahead of AI is a standalone paper that reads 1,582 answers about real estate agents from all 12 assistants in the 50 metros of the October edition: whether they name agents or brands, how much they agree, the reasons and sites behind their picks, what "affordable" does, and a plan for agents. It also has a PDF copy.
How close is the race to be AI's local pick? is a standalone paper that reads the rankings of all 1,400 markets in the October edition, built from 43,587 answers by all 12 assistants: how many answers it takes to rank at #1, #3 and #10, how far apart the top businesses are, how many assistants have to agree, and which trades are still wide open. It also has a PDF copy.
What is the AI Local Report?
The AI Local Report turns each monthly edition of AI Syndicate Local into findings people can use and cite. Each issue answers the same core questions, so issues can be compared over time:
- How often do AI assistants change the business they recommend first?
- How much do the twelve assistants agree with each other?
- Does the way a customer asks change the answer?
- Which websites do the assistants cite when they recommend a local business?
Each issue also reports what changed in the method, what the data cannot show, and what we keep private.
Where does the data come from?
Every figure comes from AI Syndicate Local. Once a month it asks 12 AI assistants (ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews, Microsoft Copilot, Siri, Alexa, Grok, Meta AI, DeepSeek and Mistral) the same three questions about each type of local business in each metro area, and ranks the businesses they name. The question wordings are fixed, so editions can be compared. The methodology page explains how answers are collected, how business names are matched, and the rule a business must clear to be ranked. Placement in the index cannot be bought.
How often is it published?
Monthly. The AI Local index is measured in the first half of each month, and the AI Local Report follows each edition. As more editions build up, the reports will add longer trends and a yearly review.
How can I cite or reuse the research?
Quote any figure with credit to AI Syndicate Research and a link to the report it comes from. Up to 50 words of text, or a single chart with its source line, may be reused that way; reproducing a report or large parts of it needs written permission. Each report carries a suggested citation at the top. Journalists and researchers who need a cut of the data that is not published can write to support@aisyndicate.com.
Sources
- AI Syndicate Local: the public index, AI Syndicate. Read Oct 5, 2026.
- AI Syndicate Local methodology, AI Syndicate. Read Oct 5, 2026.
Common questions
What is AI Syndicate Research?
AI Syndicate Research publishes AI Syndicate's measured findings on how AI assistants recommend businesses. Its main series is the monthly AI Local Report, written from the AI Syndicate Local index.
Is the research free to use?
Yes. Any figure can be quoted with credit to AI Syndicate Research and a link to the report.
Related pages.
See who AI recommends in your market.
The AI Local index is free and public: pick your state, metro and trade to see which businesses the 12 assistants name.