On October 21, 2026, this study measured the public websites of 504 small home service businesses in Ohio across six trades (plumbing, HVAC, roofing, electrical, painting and landscaping) and did not measure whether any business is actually recommended.
Of 504 domains, 441 (87.5%) returned a working homepage over HTTPS, and 34 of those 441 (7.7%) served a bot challenge instead of the page.
Of the 407 homepages used for page level checks, 390 (95.8%) had a mobile viewport tag, 328 (80.6%) had a meta description, 373 (91.6%) had a sitemap, 187 (45.9%) used LocalBusiness structured data, 149 (36.6%) named the listed city in the title, 136 (33.4%) served an llms.txt file, 5 (1.2%) blocked any of 8 AI crawlers, and 165 (40.5%) passed all 5 core checks.
Counts and percents below are the homepage results. Each row names the denominator it uses.
| Check (homepage) | Sites passing | Percent |
|---|---|---|
| Homepage loads over valid HTTPS (N=504) | 441 | 87.5% |
| Bot challenge served instead of page (N=441) | 34 | 7.7% |
| Has a title (N=407) | 397 | 97.5% |
| Title names the service (N=407) | 373 | 91.6% |
| Title names the listed city (N=407) | 149 | 36.6% |
| Mobile viewport tag (N=407) | 390 | 95.8% |
| Meta description (N=407) | 328 | 80.6% |
| Any JSON-LD structured data (N=407) | 302 | 74.2% |
| LocalBusiness type structured data (N=407) | 187 | 45.9% |
| Click to call phone link (N=407) | 327 | 80.3% |
| Contact or quote form on the homepage (N=407) | 240 | 59.0% |
| Sitemap found (N=407) | 373 | 91.6% |
| llms.txt found (N=407) | 136 | 33.4% |
| robots.txt blocks any of 8 AI crawlers (N=407) | 5 | 1.2% |
| Passes all 5 core checks (N=407) | 165 | 40.5% |
The abstract states the same gaps in round numbers: 96% are mobile ready, 81% have a meta description, 92% publish a sitemap, 1.2% block AI crawlers in robots.txt, 46% describe themselves as a local business in structured data, 37% name their city in the homepage title, 8% of reachable homepages showed a bot challenge page, about one third already serve an llms.txt file, but roughly two thirds of those appear to be generated automatically by the website builder or an SEO plugin, and 41% pass all five core checks.
Only 5 of 407 sites (1.2%) block any of the eight AI crawlers, and only 1 blocks the crawlers that power live AI search answers (OAI-SearchBot, ChatGPT-User, PerplexityBot). 29 sites (7.1%) mention an AI crawler by name, and 4 of those 29 use that to block it.
34 of 441 working homepages (7.7%) returned a challenge or captcha page to a single request instead of the website. A further 63 of 504 domains (12.5%) did not return a working homepage over HTTPS at all: 8 had a certificate that failed verification, 32 had another TLS connection error, 22 returned an error status, and 1 timed out.
74.2% of homepages carry some JSON-LD, yet only 45.9% use a LocalBusiness type. By trade, this ranges from 29.2% for electrical (N=48) and 34.3% for plumbing (N=70) to 63.9% for roofing (N=83).
91.6% of titles name the service, but only 36.6% name the city listed for the business. A broader test that also accepts Ohio or a major Ohio city name raises this to 60.9%.
33.4% of sites serve an llms.txt file. Of those 136 files, 89 (65.4%) are on a builder that appears to generate one automatically or carry a note saying an SEO plugin generated them.
The sample started from 1,055 business records collected in October 2026 from public sources, mainly OpenStreetMap and Ohio chamber of commerce directories. Records with a website were kept. Directory and social pages, national brands, and non trade organizations were removed. Each business was classified into one of six trades by category or name. Duplicate domains were removed, including two that redirected to a site already in the sample. The frame is 504 unique domains: 126 HVAC, 106 landscaping (including lawn and tree care), 100 roofing, 84 plumbing, 62 electrical and 26 painting.
For each domain the measurement made one HTTPS request to the homepage (retrying once with www only on failure), then one request each for /robots.txt, /llms.txt and /sitemap.xml. It used a normal browser user agent with a contact note, short timeouts and no crawling. All checks are on the homepage only. The measurement ran on October 21, 2026, from 4:21 to 4:27 AM Eastern Time.
A check passes only when the evidence is in the returned file. A robots.txt or llms.txt that came back as an HTML error page does not count. A sitemap counts when /sitemap.xml returns sitemap XML or robots.txt declares one. LocalBusiness schema means JSON-LD with a LocalBusiness type or a subtype such as Plumber, HVACBusiness or RoofingContractor. AI crawler blocking was tested with a standard robots.txt parser for GPTBot, OAI-SearchBot, ChatGPT-User, PerplexityBot, ClaudeBot, Google-Extended, CCBot and Applebot-Extended.
Of 504 domains, 441 returned a working homepage over HTTPS. Of those, 34 returned a bot challenge page, leaving 407 homepages for page level checks. Every percentage gives its N, and no group under 10 is reported.
For 337 of the 407 sites there was an earlier October 2026 homepage capture. Fresh and earlier results agreed for 99.7% of sites on the viewport check and 98.8% on the JSON-LD and LocalBusiness checks.
This is a convenience sample, not a random one. The sample comes mostly from prospect lists. Many records came from lists gathered to find businesses that might want website help, which may make the sample weaker than average.
It covers Ohio only, mostly Northeast Ohio, and some commercial contractors remain. Only homepages were checked, and content added by JavaScript after load is not seen, so form and schema numbers are lower bounds. The measurement network could not make plain HTTP requests, so HTTP only sites are counted with the HTTPS failures, and some TLS errors may be specific to that network. A challenge page shown here may not be shown to every AI crawler. Platform detection uses page fingerprints. Results reflect one day.
Jonathan Telep (Telep IO). 2026. AI Search Readiness of Ohio Trades Websites, October 2026. https://doi.org/10.5281/zenodo.23538029
Licensed CC BY 4.0. The DOI is the identifier for this version.
The aggregate CSV is the published data. Per site data is kept private and is not published, so no business can be identified.
The free website check reads one public page. It reports HTTPS, a mobile viewport tag, and whether common AI crawlers are blocked. It does not store the URL.