I read 31 contractor websites. Here is what was broken.

How this was counted

The count

• 7of 31Stop your phone from zooming inThe page tells the browser user-scalable=0. Someone who can't quite read your service list can't make it bigger. They don't email to complain. They go back to the search results.
• 2of 31Have a link a customer can click that does nothinghref="#" on a real, visitor-facing link. On one site it was Terms of Use, Privacy Policy and the license agreement — all three. On another, a slider button that says Click Here and doesn't. This number was wrong when this page first went up. The last section explains.
• 17of 31Give you no way to email them from the homepageNo mailto: link and no working contact form. If someone doesn't want to phone you at 9pm, there is nothing on the page for them to do. Three more have no mailto: but do have a form, so they aren't counted here.
• 14of 31Show a stale copyright year, or none at allSeven show an out-of-date year — the oldest reads 2016. Seven show no year at all. A ten-year-old date is the fastest way to make a customer wonder whether you're still trading.
• 12of 31Have a title tag Google will cut offOver 60 characters, so the end of it disappears in the search result. The longest was 100 characters. That line is the headline of your free ad on Google and half of it is missing.
• 9of 31Have no image for link previewsNo og:image. When a happy customer pastes your address into Facebook or a group chat, it renders as a gray box.
• 5of 31Print a phone number you cannot tapNo tel: link. Four of them display the number as plain text, so on a phone the customer has to memorize it, back out and dial. One had no phone number anywhere on the site at all.
• 8of 31Serve images with no alt text49 images out of 594. On the worst site, 18 of its 21 images. Screen readers announce nothing, and Google is told nothing about the photographs of your own work.
• 8of 31Have the wrong number of top-level headingsTwo pages have no h1. One has twenty-four. Search engines use that tag to work out what the page is about.
• 5of 31Have no canonical tagNothing telling Google which address is the real one, so a site reachable at more than one URL can compete against itself.
• 3of 31Have no description for search resultsTwo more have two of them, which is the same as having none. Google writes its own sentence about your business instead of the one you'd choose.
• 1of 31Has no secure version at allNo HTTPS. Every visitor gets Chrome's Not secure warning next to the address. There has never been a certificate issued for the domain. The fix is free and takes ten minutes.

The expensive ones aren't the technical ones

Nobody leaves a website because of a canonical tag. They leave because they couldn't read it, couldn't tap the number, or weren't sure the business still existed.

The zoom lock is the worst thing on this list and it's a single line of code from a template that shipped years ago. The business owner has never seen it, because it doesn't misbehave on a laptop. It only bites the customer standing in their own backyard trying to read your service list one-handed.

The second worst isn't in the count at all, because it isn't a defect — it's an omission. One of these builders has been trading since 1957 and mentions it once, in the middle of a paragraph, below the fold. Sixty-nine years in one trade in one valley is the single most convincing thing that business owns, and it is a line in a paragraph.

One site was clean

Six things the counting got wrong

This is the part that matters. A scanner will hand you a list like the one above in four seconds, and a good share of it will be false. Every one of these looked like a finding and none of them were. The sixth was mine, and it was live on this page for a day.

Six false positives, one of them mine. That is the entire difference between a scan and an audit, and it is why this is done by hand and why it takes a day rather than four seconds.

For the avoidance of any doubt: while I was doing this I also found four British spellings on my own website and shipped one of them in an email to an American business. Everybody's site has something. Mine included, twice this week.

Check your own site in ten minutes

All of this is free. None of it needs a tool.

1. Try to pinch-zoom your own homepage on your phone.If it refuses, that's the zoom lock. Ask whoever maintains the site to delete maximum-scale=1, user-scalable=0 from the viewport tag.
2. Tap your own phone number on your phone.If nothing happens, it isn't a link. Every visitor who wanted to call you had to memorize it first.
3. Scroll to your footer and read the year.If it isn't this year, fix it now. It takes thirty seconds and it's the first thing a cautious customer checks.
4. Search Google for your business name and read your own result.If the blue line trails off in an ellipsis, your title is too long. If the gray line underneath isn't a sentence you wrote, you have no meta description.
5. Paste your web address into a group chat.Whatever preview appears is what appears when a customer recommends you. If it's a gray box, you have no og:image.
6. Click every link in your own footer.Terms, privacy, careers, the social icons. On two of these sites a real one went nowhere — on one of them it was Terms of Use, Privacy Policy and the license agreement, all three.

That's seven of the twelve findings above, at no cost, in under ten minutes. If you do nothing else after reading this, do the first two.