Do AI assistants favour chains over independent local businesses?
We matched the local businesses that ChatGPT, Claude and Gemini name on 200 random AIGROW Index pages to OpenStreetMap, and set the share of chain outlets among them against the share among every business of that trade around the city: 20 countries, 22 kinds of business.
Published by Guglielmo Paesano, founder of AIGROW · working paper, 5 November 2026
- Pages
- 200
- Countries
- 20
- Kinds of business
- 22
- Named / expected
- 1.87
- Chain vs independent, same page
- 2.51
Working paper, not peer reviewed
On 200 random AIGROW Index pages, 19.8% of the 863 local businesses that ChatGPT, Claude or Gemini named were chain outlets, against 10.6% expected from the map of each city. That is 1.87 times the expected share. On the same page, a chain was 2.51 times as likely to be named as an independent.
Findings
Shares are of the businesses named by at least one of the three assistants unless it says otherwise. The tables below give every figure with its 95% interval.
AI assistants name chain outlets almost twice as often as the map predicts: 19.8% of the businesses they named, against 10.6% expected.
That is 1.87 times the expected share (1.52 to 2.50). Each assistant on its own does the same: ChatGPT names 2.02 times the expected share of chains, Claude 2.10 and Gemini 1.99. The expected share is the one the named businesses would have if each page’s were drawn from the businesses of that trade on the map around the city. 125 of the 200 pages have a named business found on the map; the others add nothing to either side.
On the same page, a chain outlet was 2.51 times as likely to be named as an independent business.
The 95% interval runs from 1.86 to 3.69. The comparison is made inside each page and then pooled, so a city full of chains does not weigh on one with few. Counting every time a business is named, not once, makes the tilt larger (2.45 times the expected share): chains are named again and again across questions and runs. A second sample of 713 pages, larger but not drawn at random and limited to businesses with a website, gives 1.95 times (1.72 to 2.22).
The tilt is strongest in gyms, hair salons, real estate agencies and car repair: in gyms 57.1% of the businesses named were chain outlets, against 10.7% expected.
Restaurants go the other way: 1.3% of the 223 restaurants named were chain outlets, against 2.6% expected. Hotels, where chains are common on the map, show a small tilt (1.25 times). In 7 of the 22 kinds of business the sampled map holds no chain at all, so there is nothing to compare. Each kind of business rests on 9 or 10 pages, so a single row shows a direction, not a size.
What it means for an independent business
When someone asks an assistant for a gym, a hair salon or an estate agent nearby, the chains start ahead. The study measures that tilt, not its cause, and a chain may simply be larger and better known. Either way, an independent has to give the assistants more to go on.
- Be the same business everywhere
- Use one name, address and phone number on your site, your Google Business Profile and the directories, so every mention an assistant reads points to the same place.
- Say what you are
- Describe your business in your page markup, with the trade, the address and the opening hours, so an assistant reading your site does not have to guess.
- Give them more than a name
- Ask your customers for reviews and keep your listings complete, so the assistants find more than a name when they look you up.
- Measure it
- Check who the assistants name for your trade in your city now, and again after any change. AI Local Fix does this groundwork for a local business, from listings and the Google profile to page markup and a review kit, then asks the same questions again and counts your closest rivals in the answers.
Named against expected
Each pair of bars sets the chain share among the businesses named against the share the map predicts for the same pages. Bars in one chart share a scale.
- Chain share, named
- Chain share expected
- Chain share, named
- Chain share expected
By assistant
Named / expected divides the chain share among the named businesses by the expected share. The last column compares a chain and an independent business on the same page.
| Assistant | Named businesses | Chain share, named | Chain share expected | Named / expected | Chain vs independent, same page |
|---|---|---|---|---|---|
| ChatGPT | 531 | 20.9% (17.7% to 24.6%) | 10.3% | 2.02 (1.62 to 2.72) | 2.81 (2.03 to 4.38) |
| Claude | 509 | 26.3% (22.7% to 30.3%) | 12.5% | 2.10 (1.66 to 3.02) | 3.05 (2.14 to 5.08) |
| Gemini | 350 | 22.0% (18.0% to 26.6%) | 11.0% | 1.99 (1.58 to 2.63) | 2.70 (1.95 to 4.03) |
| Any of the three | 863 | 19.8% (17.3% to 22.6%) | 10.6% | 1.87 (1.52 to 2.50) | 2.51 (1.86 to 3.69) |
Counting every linked name, so a business named in ten answers counts ten times, the chain share is 27.9% against 11.4% expected: 2.45 times (1.82 to 3.40).
By kind of business
Most named first. With 9 or 10 pages each, read a row as a direction; where no chain is on the map there is nothing to compare.
| Kind of business | Pages | Businesses on the map | Chain share, map | Named businesses | Chain share, named | Chain share expected | Named / expected |
|---|---|---|---|---|---|---|---|
| Restaurants | 9 | 8,276 | 2.4% | 223 | 1.3% (0.5% to 3.9%) | 2.6% | 0.52 (0.00 to 1.01) |
| Hotels | 9 | 1,962 | 30.9% | 189 | 40.2% (33.5% to 47.3%) | 32.1% | 1.25 (1.04 to 1.56) |
| Hair salons | 9 | 2,718 | 0.9% | 62 | 8.1% (3.5% to 17.5%) | 0.9% | 9.36 (4.76 to 15.74) |
| Gyms | 9 | 584 | 8.9% | 56 | 57.1% (44.1% to 69.2%) | 10.7% | 5.32 (4.06 to 7.29) |
| Real estate agencies | 9 | 328 | 6.7% | 55 | 29.1% (18.8% to 42.1%) | 7.8% | 3.71 (2.73 to 6.45) |
| Dentists | 9 | 1,056 | 0.7% | 54 | 5.6% (1.9% to 15.1%) | 0.9% | 6.25 (0.00 to 12.15) |
| Veterinarians | 10 | 240 | 3.3% | 40 | 15.0% (7.1% to 29.1%) | 6.5% | 2.32 (1.48 to 4.16) |
| Car repair shops | 9 | 1,126 | 16.8% | 39 | 53.8% (38.6% to 68.4%) | 15.1% | 3.56 (2.41 to 5.96) |
| Driving schools | 9 | 368 | 1.4% | 34 | 2.9% (0.5% to 14.9%) | 3.2% | 0.93 |
| Lawyers | 9 | 345 | 0.0% | 30 | 0.0% (0.0% to 11.4%) | 0.0% | n/a |
| Architects | 9 | 188 | 0.0% | 24 | 0.0% (0.0% to 13.8%) | 0.0% | n/a |
| Insurance brokers | 9 | 151 | 7.3% | 16 | 31.2% (14.2% to 55.6%) | 15.9% | 1.97 (0.70 to 5.79) |
| Dermatologists | 10 | 87 | 1.1% | 8 | 0.0% (0.0% to 32.4%) | 0.0% | n/a |
| Electricians | 9 | 170 | 0.6% | 7 | 0.0% (0.0% to 35.4%) | 1.6% | 0.00 |
| Accountants | 9 | 80 | 2.5% | 7 | 14.3% (2.6% to 51.3%) | 7.1% | 2.00 |
| Physiotherapists | 9 | 252 | 0.0% | 6 | 0.0% (0.0% to 39.0%) | 0.0% | n/a |
| Plumbers | 9 | 64 | 14.1% | 5 | 20.0% (3.6% to 62.4%) | 18.0% | 1.11 |
| Moving companies | 9 | 65 | 0.0% | 3 | 0.0% (0.0% to 56.1%) | 0.0% | n/a |
| Digital marketing agencies | 9 | 63 | 0.0% | 2 | 0.0% (0.0% to 65.8%) | 0.0% | n/a |
| Financial advisors | 9 | 44 | 9.1% | 1 | 100.0% (20.7% to 100.0%) | 12.5% | 8.00 |
| Cleaning services | 9 | 16 | 0.0% | 1 | 0.0% (0.0% to 79.3%) | 0.0% | n/a |
| Roofers | 9 | 15 | 0.0% | 1 | 0.0% (0.0% to 79.3%) | 0.0% | n/a |
By country
In 18 of the 19 countries with a defined ratio, the chain share among named businesses is above the expected one, and 7 have an interval wholly above 1.
| Country | Pages | Businesses on the map | Chain share, map | Named businesses | Chain share, named | Chain share expected | Named / expected |
|---|---|---|---|---|---|---|---|
| United States | 10 | 2,011 | 15.2% | 89 | 31.5% (22.8% to 41.7%) | 23.9% | 1.32 (1.06 to 2.87) |
| Canada | 10 | 939 | 5.1% | 76 | 28.9% (20.0% to 40.0%) | 12.2% | 2.37 (1.53 to 8.96) |
| Portugal | 10 | 1,971 | 4.1% | 75 | 6.7% (2.9% to 14.7%) | 8.0% | 0.84 |
| Denmark | 10 | 1,485 | 7.3% | 74 | 13.5% (7.5% to 23.1%) | 12.3% | 1.10 (0.60 to 4.30) |
| United Kingdom | 10 | 762 | 5.8% | 70 | 17.1% (10.1% to 27.6%) | 11.9% | 1.44 (1.13 to 3.04) |
| Germany | 10 | 2,234 | 4.6% | 64 | 14.1% (7.6% to 24.6%) | 9.0% | 1.56 (0.00 to 1.68) |
| Netherlands | 10 | 1,092 | 2.6% | 56 | 8.9% (3.9% to 19.3%) | 2.8% | 3.17 (0.00 to 6.79) |
| Australia | 10 | 1,216 | 6.2% | 49 | 38.8% (26.4% to 52.8%) | 14.3% | 2.72 (1.93 to 10.10) |
| France | 10 | 1,613 | 7.6% | 42 | 28.6% (17.2% to 43.6%) | 10.3% | 2.76 (0.83 to 7.39) |
| Belgium | 10 | 908 | 2.0% | 42 | 14.3% (6.7% to 27.8%) | 2.5% | 5.81 (2.24 to 8.13) |
| Ireland | 10 | 321 | 4.4% | 41 | 14.6% (6.9% to 28.4%) | 10.6% | 1.38 (0.59 to 4.20) |
| Italy | 10 | 1,013 | 2.0% | 40 | 2.5% (0.4% to 12.9%) | 1.5% | 1.66 (0.00 to 28.00) |
| Mexico | 10 | 410 | 14.6% | 26 | 38.5% (22.4% to 57.5%) | 14.4% | 2.67 (2.01 to 12.33) |
| Poland | 10 | 627 | 4.9% | 23 | 8.7% (2.4% to 26.8%) | 1.8% | 4.78 |
| Sweden | 10 | 141 | 8.5% | 23 | 21.7% (9.7% to 41.9%) | 9.7% | 2.24 (0.00 to 4.39) |
| Austria | 10 | 601 | 5.2% | 19 | 15.8% (5.5% to 37.6%) | 10.7% | 1.47 (0.00 to 11.50) |
| Spain | 10 | 357 | 3.9% | 18 | 44.4% (24.6% to 66.3%) | 5.1% | 8.64 (3.17 to 11.77) |
| Switzerland | 10 | 155 | 9.0% | 16 | 37.5% (18.5% to 61.4%) | 15.2% | 2.46 |
| Norway | 10 | 236 | 5.1% | 15 | 13.3% (3.7% to 37.9%) | 7.0% | 1.92 |
| Brazil | 10 | 106 | 0.0% | 5 | 0.0% (0.0% to 43.4%) | 0.0% | n/a |
Other ways to draw the line
Every variant keeps the tilt, from 1.49 to 3.41 times the expected share, and every interval stays above 1.
| Variant | Named businesses | Chain share, map | Chain share, named | Chain share expected | Named / expected | Chain vs independent, same page |
|---|---|---|---|---|---|---|
| Main rule | 863 | 6.3% | 19.8% | 10.6% | 1.87 (1.52 to 2.50) | 2.51 (1.86 to 3.69) |
| Strict rule: a Wikidata brand only | 863 | 4.8% | 17.5% | 8.2% | 2.14 (1.66 to 3.05) | 2.89 (2.04 to 4.51) |
| Loose rule: any brand or operator tag | 863 | 9.6% | 24.4% | 14.1% | 1.73 (1.47 to 2.18) | 2.29 (1.80 to 3.07) |
| No tags: the same name at two places at least 500 m apart | 863 | 3.3% | 12.1% | 3.5% | 3.41 (2.61 to 4.39) | 4.32 (3.03 to 6.07) |
| Each brand counted once per page | 855 | 4.5% | 19.1% | 8.3% | 2.29 (1.88 to 2.98) | 3.20 (2.44 to 4.60) |
| Without hotel pages | 674 | 3.3% | 14.1% | 4.6% | 3.08 (2.41 to 3.90) | 4.53 (3.11 to 6.60) |
| Without car repair pages | 824 | 5.6% | 18.2% | 10.4% | 1.75 (1.43 to 2.30) | 2.28 (1.72 to 3.29) |
| Businesses with a website only | 578 | 9.9% | 21.8% | 14.6% | 1.49 (1.24 to 1.93) | 1.92 (1.42 to 2.85) |
The names that did not link
The worst case: every unlinked name counts as independent unless it carries a chain brand or operator found on the same trade’s pages, and a name that fits several businesses counts as a chain only when all of them are.
| Assistant | Specific names | Not linked | Chain share, at least | Chain share expected |
|---|---|---|---|---|
| ChatGPT | 6,888 | 4,834 | 9.8% (9.2% to 10.6%) | 5.5% |
| Claude | 8,895 | 6,788 | 9.5% (8.9% to 10.1%) | 6.1% |
| Gemini | 7,076 | 5,461 | 6.2% (5.7% to 6.8%) | 5.2% |
| Any of the three | 22,859 | 17,083 | 8.6% (8.2% to 8.9%) | 5.6% |
The margin is smallest for Gemini: 6.2% against 5.2%.
A second, larger sample
713 Index pages in 8 countries whose map holds only businesses that list a website. Not drawn at random, so a replication, not the main result.
| Assistant | Named businesses | Chain share, named | Chain share expected | Named / expected | Chain vs independent, same page |
|---|---|---|---|---|---|
| ChatGPT | 1,665 | 30.3% (28.1% to 32.5%) | 14.8% | 2.05 (1.79 to 2.35) | 3.36 (2.72 to 4.18) |
| Claude | 1,608 | 35.1% (32.8% to 37.4%) | 16.6% | 2.11 (1.84 to 2.46) | 3.72 (2.96 to 4.81) |
| Gemini | 1,116 | 30.8% (28.2% to 33.6%) | 15.2% | 2.03 (1.73 to 2.38) | 3.25 (2.51 to 4.20) |
| Any of the three | 2,563 | 28.4% (26.7% to 30.2%) | 14.5% | 1.95 (1.72 to 2.22) | 3.12 (2.55 to 3.84) |
On the 670 pages outside the random sample the ratio is 1.96 (1.72 to 2.26), and in each of its 8 countries the interval is wholly above 1.
How the study was made
- The answers
- The November 2026 edition of the AIGROW AI Visibility Index asked ChatGPT, Claude and Gemini the same questions about one trade in one city, several times over. 7,198 answers on the pages studied came back. No assistant was asked anything new for this study.
- The pages
- 200 local Index pages drawn at random, 10 in each of 20 countries, covering 22 kinds of business in 88 cities.
- The map
- Every OpenStreetMap object of the page’s trade within the page’s radius, with its brand and operator tags: 18,198 businesses on the 200 pages. Branches with the same name within the radius count as one business. Names made only of trade and city words are left out, since an answer cannot single them out.
- Chain or independent
- A business is a chain outlet when the map ties it to a brand with a Wikidata entry, to a brand name shared with other places, or to a private operator that runs places under different names. Public bodies do not count as operators. The rule was fixed on 5 November 2026, before any figure about named businesses was computed. It finds 1,141 chains among the 18,198 businesses on the map, 881 of them through a Wikidata brand, and 171 among the 863 named.
- From a name to a business
- A name counts when it has a distinctive word beyond the trade and the city, and it is linked when it fits exactly one business on the page’s map: 5,580 of 22,859 such names. 196 fit more than one business and stay unlinked. A named business counts once, however often it is named.
- The expected share
- The chain share the named businesses would have if each page’s named businesses were drawn from that page’s map. Each page weighs as much as the businesses named on it, so the expected share differs from the pooled share of chains on the map, 6.3% of all 18,198 businesses.
- Statistics
- Shares of businesses carry 95% Wilson intervals. Every other interval is a bootstrap of whole pages (2,000 resamples), which keeps the mix of pages in the uncertainty; when more than one resample in ten has no defined value, for example no chain on the map, the interval is left out. The same page comparison is a Mantel-Haenszel risk ratio with the page as stratum.
Limits
- The map is not the street
- OpenStreetMap is drawn by volunteers, and chain outlets are often added in bulk from brand lists, so the map likely misses more independents than chains; by how much was not measured. Missing independents raise the share of chains on the map, which makes the tilt look smaller, and cannot be linked to the answers, which makes it look larger. The worst case above covers the second.
- Most names do not link
- 74.7% of the specific names have no single match on the map: some businesses are missing from it, some names are spelled differently, and some do not exist. Even if every one of them were an independent business unless it carries a chain brand, chains would be at least 8.6% of the names against 5.6% expected. The tilt does not depend on these names, but its size does.
- Tags vary
- Brand tagging differs by country and trade; the strict, loose and tag free rules bracket that. In car repair a workshop tagged with the one car make it serves counts as a chain outlet, since it is usually a franchised dealer; without car repair pages the ratio is 1.75.
- Small groups
- Each kind of business and each country rests on 9 or 10 pages of the random sample. Read a single row as a direction, not a size.
- The second sample is not random
- The 713 pages cover the cities where AIGROW had already collected map data, in 8 countries, and their map holds only businesses that list a website. They are a replication, not the main result.
- Written knowing earlier work
- An earlier AIGROW Labs analysis of the same pages had already seen chains named more often, and the rule was written knowing that. It was fixed before any figure about named businesses was computed, and the strict, loose and tag free rules agree.
- What chain means here
- A business that shares a brand or a private operator with other places. Franchises and company owned chains are not told apart.
- Cause
- The map counts businesses, not their size, footfall, reviews or presence on the web. A chain gym may simply be larger and better known. This study measures the tilt, not its cause.
- One edition
- All answers come from the November 2026 edition of the Index.
- Competing interests
- The author founded AIGROW, which publishes the Index and sells services that aim to improve how businesses appear in AI answers. The study had no external funding.
Rankings from the AI Visibility Index
See which businesses and products ChatGPT, Claude, Gemini and Grok named for the categories and places below.
How to cite
Guglielmo Paesano, Chains and independents in AI answers about local businesses, AIGROW Labs working paper, 5 November 2026, aigrow.gg/research/ai-chains-vs-independents-2026.
A working paper, not yet peer reviewed. The figures on this page are the paper’s.
Answers: AIGROW AI Visibility Index, November 2026 edition, DOI 10.5281/zenodo.23934866.
Map data © OpenStreetMap contributors, ODbL.
The study names no business. The files that link names to businesses stay with AIGROW Labs.
Is an assistant naming a chain instead of you?
The free scan asks ChatGPT, Claude, Gemini and Grok your own customers' questions, shows who they name instead of you, and checks whether their crawlers can read your site.
AIGROW Labs · 5 November 2026