Original research · 28 August 2026

We checked 31 local service business websites. Eight of them had no working website at all.

Salons, plumbers, roofers, auto shops and nail bars across seven US states. We ran every one through our own website checker, then verified the results by hand a day later. This page is the whole result, including the parts that make the method weaker.

How we picked the 31

This was not a random sample, and that matters for every number below.

The 31 businesses came out of our own prospect database, which is built by scraping Google Maps listings for specific trades in specific cities. From roughly 6,000 entries we took the ones that had a website listed, a published email address, and no salesperson already assigned to them. That left 31.

By state: Arizona 12, Georgia 5, Michigan 5, Tennessee 3, Florida 2, North Carolina 2, Texas 2. The trades are mixed: hair and nail salons, plumbing, roofing, HVAC, auto repair, handyman services, one games supplier.

We are not naming the businesses. They did not ask to be looked at, and a public list of companies whose website is broken would hurt them for no good reason. The aggregate numbers are the useful part.

How we checked

Every domain went through our own checker on 27 August 2026. It reads what the site already shows any visitor: HTTP response, TLS certificate, security headers, DNS records for mail, page title and headings, meta description, structured data, sitemap, mobile viewport and the required legal pages. It does not execute JavaScript.

On 28 August we re-checked reachability separately, with a normal browser user agent, over both HTTPS and HTTP, with DNS resolution as a first step. That second pass is what the reachability numbers below are based on.

Finding 1: a quarter of them had no working website

Eight of 31, or 26 percent, had no website a customer could reach. Seven of those domains do not resolve in DNS any more, which usually means the registration lapsed or was never renewed. One resolves but its homepage returns a 404 error.

All eight are still listed on Google Maps with a phone number and an address, and the website field still points at the dead domain. Anyone who finds the business and clicks through lands on an error page or a parked domain.

One of the eight is a special case worth naming as a pattern rather than a company: the business had moved to a new domain, and the old one is dead. The new site works fine. The old address is still what several directories hand out.

Finding 2: the sites that do work are mostly fine

The 19 reachable sites scored between 29 and 95 out of 100, with a median of 79 and a mean of 75. That is not a picture of an industry in ruins. Most of these websites do their job.

Eleven of the 19 had at least one finding we classify as critical. Two sites had no working mobile layout. One had no page title at all, which means its Google result shows nothing clickable.

Finding 3: email spoofing protection is mostly missing

This was the most common serious problem, and it is invisible from the website itself.

In plain terms: for five of these businesses, anyone on the internet can send an email that looks like it came from their domain. An invoice with the wrong bank details, sent to one of their customers, would arrive looking legitimate. Each of these is one DNS record, and none of them cost money.

Finding 4: 13 percent blocked us, and a careless audit would call them broken

Four of the 31 sites returned HTTP 403 to an automated request. They are behind bot protection. Opened in a normal browser they work perfectly.

We are reporting this because it is the failure mode of every automated website audit, including ours. Our first pass on 27 August flagged one of these as unreachable and it was not. If you receive an audit telling you your site is down, check it yourself in a browser before you believe it. We excluded all four from the scored results rather than guess.

Finding 5: none of them are readable by AI assistants

Every one of the 19 sites had its address, opening hours and services written on the page for a human to read, and none of them had it in a format a machine reads reliably. When somebody asks an assistant for a plumber nearby, the assistant has to guess, and guessing usually means leaving that business out.

We want to be careful here. There is no published evidence from Google, OpenAI, Anthropic or Perplexity that any particular markup type makes an AI assistant more likely to cite you. Google says explicitly that no special AI markup is required. What structured data does is remove ambiguity, and 17 of 19 of these sites are ambiguous.

What this does not show

• It is not a random sample. These 31 came from a prospect list built for selling websites, filtered to businesses that had an email address published. That selection is biased in ways we cannot measure.
• Thirty-one is a small number. The 26 percent figure could easily be 15 percent or 40 percent in a larger sample.
• We used our own tool. Another checker would score these sites differently.
• It is one point in time, from one place. A site that was down when we looked may have been up an hour later.
• We cannot tell a closed business from a lapsed domain. Some of the seven dead domains may belong to companies that no longer trade.

The one number we would repeat

If you take one thing from this: the most common serious problem was not a slow or ugly website. It was a business whose website had quietly stopped existing.

Nobody told them. The Google listing still points at it. The phone still rings, so nothing feels wrong. The only people who notice are the customers who click the link and leave.

S7 Digital builds websites for local service businesses in the United States. We do not sell ongoing SEO or paid advertising. We ran this study on our own prospect list, with our own tool, and we have an obvious interest in you concluding that websites matter. The raw method is above so you can decide how much weight the numbers deserve.