The headline table
Seven industries, one table.
One row per industry index, computed from the same data files each report renders from. Named rate is the share of the crawled, resolved-domain homepages that AI named in its own city answer, so it is comparable across industries even though each list of businesses AI named is a different size.
| Industry | Cities | Sites attempted | Sites read | Named by AI | Named rate | |
|---|---|---|---|---|---|---|
| Med spas | 245 | 1,531 | 1,352 | 1,116 | 0.3% | Read the index |
| Roofers | 60 | 569 | 520 | 277 | 0.3% | Read the index |
| Dentists | 40 | 435 | 366 | 134 | 0.2% | Read the index |
| Law firms | 60 | 875 | 685 | 297 | 0.3% | Read the index |
| Chiropractors | 40 | 602 | 460 | 192 | 0.2% | Read the index |
| HVAC | 60 | 577 | 506 | 299 | 0.3% | Read the index |
| Plumbers | 60 | 443 | 409 | 305 | 0.3% | Read the index |
| Total | 565 | 5,032 | 4,298 | 2,620 | 0.3% |
Sites attempted and sites read are from each industry's own homepage crawl (crawled September 10, 2026 for med spas, September 11, 2026 for roofing, September 11, 2026 for dentists, September 11, 2026 for law firms, September 12, 2026 for chiropractors, September 12, 2026 for HVAC and September 15, 2026 for plumbers). Named by AI is the total business names across every city answer in that industry's index, a different and usually larger population than the named, resolved-domain sites the crawl could read.
Related: Study 02, med spa pricing transparency reads what kind of price the med spa row above actually publishes, by state and by city.
Related: Study 03, where AI gets its answers about local businesses reads the citation behind every named business above: the business's own site, its Google profile, or a directory, and how that split changes by industry.
Get your business recommended by AI in 90 days.
Per industry
What each index found.
Two strongest findings per industry, each one read straight off that index's own crawl and city answers.
Med spas
245 cities checked, 1,352 homepages read, 1,116 business names across every city answer.
No structured data on the homepage. 331 of the 1,531 sites we screened carry no schema.org markup at all.
The answers are still open. Google AI Mode named 341 of the 1,352 homepages we crawled, and no single clinic was pulling ahead in most of the cities we checked.
Source: STACK-FINDINGS-2026-09-10.md. Read the full index
Roofers
60 cities checked, 520 homepages read, 277 business names across every city answer.
The manufacturer badge does not sort the two groups. A certification tier is the badge this trade sells itself on. 26% of the roofing companies the answer named claim one, and 25% of everyone else claims one too. On this day, in this market, it told the two groups apart by one point.
Most of the markup describes the website, not the business. 85% of the homepages we read carry structured data. 38% use it to say the business is a roofing contractor, and 26% carry only the website record the theme ships. Naming the business type is a one line change on most sites.
Source: ROOFING-INDEX-FINDINGS-2026-09-11.md. Read the full index
Dentists
40 cities checked, 366 homepages read, 134 business names across every city answer.
The letters after the name do not sort the two groups. DDS or DMD after a name is the badge this trade sells itself on. 37% of the practices the answer named print it on the homepage, and 48% of everyone else prints it too. Any provider credential at all reads the same way, 44% against 54%. An Invisalign provider tier is claimed on 0% of the sites we read.
The named practices name more of the work they do. Invisalign or clear aligners offered: 76% of the named homepages say so, against 51% of everyone else. Emergency dentistry offered: 70% against 52%. Writing down every treatment you already do is the cheapest move on this page.
Source: DENTAL-INDEX-FINDINGS-2026-09-11.md. Read the full index
Law firms
60 cities checked, 685 homepages read, 297 business names across every city answer.
There is no empty city in this market. Google AI Mode named at least 3 personal injury firms in every one of the 60 cities we checked, 297 firms in all. The question is already answered here, with about 5 firms in each answer, which makes this a different job from a market where the AI still has nothing to say.
Being listed reads differently from being picked. The listings and memberships this trade buys sit on the same number in both halves of the comparison: Avvo 16% against 14%, American Association for Justice 13% against 11%, Lead Counsel or Expertise badge 8% against 8%, Justia profile 7% against 7%, named firms against everyone else. The selective awards do separate: Million Dollar Advocates Forum 22% against 12%, Multi-Million Dollar Advocates Forum 15% against 6%, National Trial Lawyers Top 100 25% against 16%, Best Lawyers 18% against 10%. A profile and an award are not the same kind of thing, and this data reads them differently.
Source: LAW-FIRMS-INDEX-FINDINGS-2026-09-11.md. Read the full index
Chiropractors
40 cities checked, 460 homepages read, 192 business names across every city answer.
Every city already has an answer. Google AI Mode came back with practice names in every one of the 40 cities we asked, 192 names in all. The list is short every time: about 4.8 names per city, never more than 7. The job in this market is joining a list that already exists.
The named practices write more and list more. A blog or articles section is linked from 67% of the named homepages and 44% of everyone else. Knee, shoulder or joint pain sits on 57% against 40%. The named homepages run a median of 1,135 words to 821. Nothing in the technology stack comes close to those distances.
Source: CHIROPRACTIC-INDEX-FINDINGS-2026-09-12.md. Read the full index
HVAC
60 cities checked, 506 homepages read, 299 business names across every city answer.
Every city already has an answer. Google AI Mode came back with company names in every one of the 60 cities we asked, 299 names in all. The list runs 3 to 8 names per city, about 5 on average. The job in this market is joining a list that already exists.
The named companies write more and list more. A blog or articles section is linked from 72% of the named homepages and 51% of everyone else. Financing offered sits on 84% against 68%. The named homepages run a median of 1,202 words to 1,047.
Source: HVAC-INDEX-FINDINGS-2026-09-12.md. Read the full index
Plumbers
60 cities checked, 409 homepages read, 305 business names across every city answer.
Fewer than half of the named plumbers even had a findable website. Google AI Mode named 305 plumbing companies across the 60 cities (one individual freelancer excluded as not a company). Only 144 of them, 47 percent, could be matched to a real website at all: 105 were already in our screened list and 39 more were found by a verified hostname guess. The other 161 have no domain on record. A plumbing company's name is built almost entirely from category words, "plumbing", "plumber", "drain", "rooter", "water", "service", so a guessed hostname usually has nothing distinctive left to test, and 47 percent is the lowest resolution rate of the six trades this series has run, below HVAC at 57, chiropractic at 55, law firms at 63, dentistry at 69 percent.
126 named sites were actually read and counted. Of the 144 resolved domains (143 unique, one business named twice under two names in the same city), all were crawled. 126 were read and counted, 7 sat behind a bot wall, 9 failed for another reason and 1 was an unreadable stub. An audit then tested every resolved row for a name match, a city match and enough plumbing content: 23 failed the letter of that test, 8 turned out to be real wrong-domain matches and were blanked, and 15 were kept with the evidence written down because the failure was in the audit's own name-matching (an apostrophe, an ampersand, a franchise page that never repeats the named city) rather than a wrong site. Every named-versus-unnamed comparison on this page rests on these 126 sites, about 41 percent of the 305 companies the answer actually named.
Source: PLUMBING-INDEX-FINDINGS-2026-09-15.md. Read the full index
Named versus unnamed, across industries
What the named businesses do differently.
A signal is compared here only when its exact label appears in at least four of the six industry crawls. This is a same-day comparison, not a cause: named businesses were named for reasons this crawl cannot see.
The 5 widest, most consistent gaps
- Google Tag Manager installed+12.5 pts (6/6)
- A blog or news section linked+9.2 pts (4/6)
- A review count shown on the homepage+7.2 pts (5/6)
- Copyright year 2026 in the footer+6.2 pts (6/6)
- A click to call phone link+4.6 pts (5/6)
3 signals with no real gap (within 3 points)
- Https0 pts (5/6)
- Says 'near me' somewhere on the homepage+0.3 pts (6/6)
- FAQPage schema specifically-0.7 pts (6/6)
Reading across the top five: named businesses are not winning on any single silver bullet. They are winning on basic marketing hygiene (a tag manager actually installed), on a site that gets touched at least once a year (a current copyright year), and on a homepage that gives an answer engine more to quote (a review count, a blog, a click to call link). Table stakes checks that almost everyone already has, like https and a current analytics tag, do not separate the two groups at all. Same-day comparison, not a cause.
Source: KKV Labs, cross-industry comparison of the six stack crawls above, built 2026-09-15. Full detail in NAMED-VS-UNNAMED-FINDINGS-2026-09-15.md and named-vs-unnamed-2026-09.json.
Read the full study: all 12 largest gaps, all 8 signals with no real gap, and the per-industry breakdowns behind this section.
Method
How this report is built.
Each industry gets its own index first: a fixed buyer question asked in Google AI Mode, from a fresh logged out session, once per city, and the business names it gave recorded as a dated snapshot.
The morning after, we crawl every candidate homepage plus every AI-named business we could resolve to a domain, reading the same technology and content signals for all of them.
A business counts as named only when its answer-engine listing resolves to a homepage we could read, which is why named by AI (the table above) and named, resolved sites (the rate) are different numbers.
This report adds nothing new to that data: it reads the same six data files each industry's own report renders from and rolls them into one table and one cross-industry comparison.
The cross-industry comparison only includes a signal when its exact label text appears in at least four of the six industry crawls, since each crawl defines its own signal set.
Every comparison on this page, in every industry index, is a same-day reading and not a cause. None of it is a ranking, an endorsement, or a judgement of anybody's work.
Changelog
What has changed.
- 16 Sep 2026: plumbers added, seven industries.
- 15 Sep 2026: first edition, six industries.
- This report is refreshed monthly, and grows as more industries are added.
Cite this report
KKV Labs (2026). AI Local Visibility Report, September 2026 edition. KKV Labs LLC. https://www.kkvlabs.com/research/ai-local-visibility-report-2026
Dataset DOI: https://doi.org/10.5281/zenodo.22765584 (all versions: https://doi.org/10.5281/zenodo.22765583)
GitHub: dataset and crawl code. Licence: CC BY 4.0.
Data
565 US cities across seven industries, crawled 10 to 16 Sep 2026 (seven crawls, one per industry), 4,298 homepages read. Columns: industry, cities, sitesAttempted, sitesRead, namedByAi, namedRate.
Licence: CC BY 4.0. Browse the datasets on GitHub DOI 10.5281/zenodo.22765584
Questions
What people ask about this report.
What is the AI Local Visibility Report?
Which industries does it cover, and which is next?
What do the named businesses actually do differently?
Can I cite these numbers?
Next step
We build one industry at a time, and publish a report only once we have real numbers behind it.
Get your business recommended by AI in 90 days.
We run the 10 questions your customers ask ChatGPT, Perplexity and Google AI Mode in your city and show you who gets named today. Then we fix the pages, listings and answers that decide it.
$1,000 a month, 3-month minimum as it takes time to show results. If AI is not naming you by day 90, you get every dollar back.
One business per industry per city. Founder led. Dated proof every 30 days.