Plumbing websites in 2026: do they load, and can a customer book?

Chat and booking: a phone-first trade

Plumbing leads come by phone. That shows up in the data: 94.3% of 70 sites show no chat widget, against 89.2% across all niches (n=738), and 78.6% show no chat and no booking option, against 64.1%.

Across the 623 home-service sites in the scan, 45 linked at least one of the platforms we looked for: Angi on 34, ServiceTitan scheduler on 18, Housecall Pro on 8 and Thumbtack on 6 (a site can link more than one). For plumbing, 58 of 59 sites linked none; 56 of those also had no chat. Plenty of plumbers take bookings through office software that never appears on the public site, so this is an upper bound.

Website health

Of 230 plumbing domains checked from 9 to 13 November 2026, 156 (67.8%) loaded a normal homepage.

Dead domains were 13.0%, below the all-niche 16.6%. Visibly broken homepages were 7.0% (16 domains), above the all-niche 5.5%. Bot walls on 22 domains (9.6%) are not counted as broken, but they matter if you rely on tools that visit your site.

What plumbers publish for machines

Access is open: 2 of 91 reachable sites block any AI crawler, and none block ChatGPT-User. Labels are the gap. 30 sites (33.0%) carry no JSON-LD, and 28.6% describe themselves as a local business. llms.txt is served by 39.6% of sites, above the all-niche 30.5%, but we did not test whether any AI system reads those files.

On basics, 56.0% of 91 sites have a title tag of 30 to 65 characters, 72.5% have an H1 and 83.5% have a meta description.

Key findings in short

• 94.3% of 70 sites have no chat widget; 78.6% have neither chat nor booking.
• 98.3% of 59 sites link no booking or lead platform.
• 67.8% of 230 domains load normally; 13.0% do not resolve.
• 33.0% of 91 reachable sites have no JSON-LD; 8.8% have FAQPage markup.
• 2.2% of 91 reachable sites block at least one AI crawler.

What this means for a plumbing company

1. Make the phone number a tap-to-call button that stays visible on mobile, then add a short form for non-urgent jobs (water heater quotes, fixture swaps) so after-hours visitors still leave details.
2. Show your license number and service area on every page; both are trust signals and both belong in your markup.
3. Write separate pages for the jobs people search for: drain cleaning, water heater replacement, sewer line repair, repiping. One page per job, one clear next step.
4. Add LocalBusiness markup with an area-served list. Only 28.6% of reachable sites have it. The schema markup generator helps.
5. If your scheduling software offers a customer-facing booking link, put it on the site and confirm it works on a phone.

Help available: website design for plumbers and SEO for plumbers. See also the AI search readiness study and the research index.

Method and limits

Our crawler visited on 8 November 2026 and asked each site for its homepage, robots.txt and llms.txt one time, openly. We sampled 120 plumbers from a US business list; 91 answered with HTTP 200 and form the base for every share in the AI tables.

The health sample is 230 domains, each looked up once and requested once from 9 to 13 November 2026 (latest check kept). The broken group covers 404 homepages, SSL failures, server errors, parked or for-sale pages, spam hijacks, default host pages, suspended accounts, WordPress critical errors and PHP errors; bot walls (401, 403, 429) are not part of it.

On 9 November 2026 we read the home page plus up to 7 internal pages of each site (n=70) for chat widgets and booking options. A second pass (n=59) added the careers page and searched for Housecall Pro, Jobber, Angi, Thumbtack and ServiceTitan signatures. Companies may use these tools without linking them, which makes no-partner shares upper bounds. We never spoofed a bot.

The data cannot show whether any AI system reads or cites these sites, how much traffic they get or how they rank. The samples are not random draws of every business. At the 95 percent confidence level a share on 91 sites can be off by up to about 10 points, and on the 230-domain health check by about 6. Treat close gaps with the all-niche numbers as noise.