Kodinav research · October 2026
The state of US small-business websites, 2026
We tested the websites of 1,186 independent shops, restaurants, dentists, clinics and offices in Atlanta, Chicago, Denver, Houston, Los Angeles, New York, Phoenix and Seattle. Most load fine on a phone, but 78% fail automated accessibility checks against the standard ADA lawsuits measure sites by, and half make a customer copy the phone number by hand.
- 78%
- fail automated WCAG accessibility checks
- 52%
- have text with too little contrast
- 52%
- have no tap-to-call link
- 15%
- of listed websites no longer work
What we found
01
78%
Nearly four in five fail automated accessibility checks
Free tool: website accessibility checker →We ran axe-core, the open-source engine behind most accessibility testing tools, against every working homepage for WCAG 2.1 Level A and AA rules, the standard courts and settlements use. 78% had at least one failure; the typical site failed 2 different rules.
Automated checks find only part of the problems, so the real share of sites with barriers is higher, not lower.
02
52%
- Text with too little contrast52%
- Links with no accessible name39%
- Images with no text alternative18%
- Embedded frames with no title11%
- Zoom blocked on phones11%
- Buttons with no accessible name9%
The most common failures are quick to fix
Free tool: what WCAG 2.1 AA means in practice →Low-contrast text tops the list, then links with no name (usually icon links to Instagram or Facebook) and images with no text alternative. Each is a few lines of code or a colour change.
03
11%
- Sites loading an accessibility overlay widget11%
- Of those, still failing automated checks78%
One in ten uses an overlay, and most still fail
Free tool: do accessibility overlays work? →11% of sites load an accessibility overlay (a toolbar widget sold as one-line compliance). Of those 96 sites, 78% still fail automated WCAG checks, about the same as sites without one. Overlays change how a page looks on request; they don't fix the code a screen reader reads.
04
52%
Half make a customer copy the phone number by hand
Free tool: free website audit →Only 48% of homepages have a tap-to-call link, and 23% show a way to book online. On a phone, a number written as plain text means select, copy, switch apps and paste, and many customers call the next business instead. Dentists and clinics do best, at 70% and 73%.
05
15%
One in 7 listed websites no longer works
Free tool: check if a website is down →Of 1056 websites we could test, 7% point to a domain that no longer exists, and the rest are parked or for-sale pages, server errors or sites that never answer. Some businesses will have closed, but a dead link on a map listing sends a live customer to a competitor.
06
70%
- No business details for Google (LocalBusiness schema)70%
- No share image for link previews38%
- No search description34%
Most leave Google to guess their hours and address
Free tool: schema markup generator →70% don't give search engines their business type, address and opening hours in machine-readable form, which feeds local results and AI answers. 34% have no search description and 38% no share image.
07
59/100
- Score under 50 on mobile21%
- Main content takes over 4 seconds88%
- Score 90 or more5%
On a phone, the typical site is slow
Free tool: website speed test →In a random sample of 100 sites run through Lighthouse's mobile test (the engine behind Google's PageSpeed Insights, which simulates a mid-range phone on a slow 4G connection), the median score was 59/100 and the main content took 9.2 seconds to appear; Google counts more than 4 seconds as poor. The median page weighed 3.1 MB.
By type of business
Share of working websites in each group. * Fewer than 30 sites: read as indicative only.
| n | Fail WCAG checks | Low contrast | Unnamed links | Tap to call | Online booking | Search description | |
|---|---|---|---|---|---|---|---|
| Gyms, studios & trades | 78 | 82% | 62% | 42% | 46% | 14% | 71% |
| Dentists | 30 | 90% | 69% | 62% | 70% | 37% | 73% |
| Restaurants, cafés & bars | 394 | 69% | 41% | 35% | 44% | 25% | 59% |
| Clinics & health | 79 | 89% | 71% | 56% | 73% | 29% | 76% |
| Hotels & motels | 7* | 57% | 43% | 14% | 71% | 71% | 86% |
| Offices & professional services | 61 | 89% | 67% | 39% | 56% | 8% | 75% |
| Shops & retail | 254 | 84% | 53% | 39% | 42% | 23% | 68% |
| All | 903 | 78% | 52% | 39% | 48% | 23% | 66% |
By city
About 110 working sites per city, so differences of a few points between cities are within the margin of error.
| n | Not working | Fail WCAG checks | Low contrast | Tap to call | Overlay | |
|---|---|---|---|---|---|---|
| Atlanta | 112 | 15% | 82% | 58% | 52% | 11% |
| Chicago | 109 | 14% | 69% | 47% | 41% | 7% |
| Denver | 112 | 18% | 69% | 49% | 49% | 8% |
| Houston | 107 | 18% | 86% | 58% | 52% | 7% |
| Los Angeles | 117 | 11% | 78% | 43% | 55% | 24% |
| New York | 110 | 18% | 82% | 57% | 47% | 12% |
| Phoenix | 108 | 18% | 78% | 58% | 53% | 11% |
| Seattle | 128 | 5% | 79% | 45% | 38% | 6% |
What they're built on
Share of the 903 working sites by the platform we could detect. "Custom or other" includes hand-built sites and platforms we don't fingerprint.
- Custom or other29%
- WordPress25%
- Squarespace16%
- Shopify9%
- Wix9%
- Next.js3%
- GoDaddy builder3%
- Webflow2%
- Weebly2%
- Google Sites2%
Five checks for your own site
- Every image has alt text, and every icon link and button has a name a screen reader can say.
- Text has enough contrast against its background (4.5:1 for normal text).
- Your phone number is a tap-to-call link on every page.
- Google has your business type, address and hours in schema markup.
- The domain renews automatically, to an email address someone reads.
See where your site stands
The free accessibility checker scans your homepage for the common WCAG failures in seconds, each explained with how to fix it.
Check my website →How we did it
The list. Shops, restaurants, cafés, bars, dentists, clinics, hotels, offices, studios and gyms with a website tag on OpenStreetMap in Atlanta, Chicago, Denver, Houston, Los Angeles, New York, Phoenix and Seattle, exported in October 2026. We removed chains (anything carrying a brand tag, or a domain used by more than two locations), .gov, .edu and .org sites, and listings that point only to a social page. From the 30,977 businesses left, we drew a seeded random sample of 150 per city.
The visit. One visit per site from our research crawler (user agent "KodinavResearch"), which honours robots.txt: 7 sites opted out and 123 blocked automated visits; both are left out. Each homepage was fetched over https and http, then loaded in a mobile browser at 390 pixels wide, where axe-core 4 checked it against WCAG 2.0 and 2.1 Level A and AA rules. Sites with no answer were retried once.
The measures. "Fails accessibility checks" means axe-core reported at least one WCAG A or AA violation on the homepage. "No longer works" means the domain does not resolve, the server does not answer, it returns an error, or it shows a parking page. Contact, booking and search measures are read from the homepage only. Speed comes from Lighthouse 12 with its standard mobile settings, run on a seeded random sample of 100 working sites.
The limits. OpenStreetMap lists the businesses volunteers have mapped, so this is a sample, not a census, and leans toward businesses someone took the trouble to map. Automated checks find only part of accessibility problems, and a failed check is not a legal finding. We publish only totals and never name a business.
The data
Cite this study
Kodinav (2026). The state of US small-business websites, 2026. https://kodinav.com/research/us-small-business-websites-2026