Everyone in our trade has an opinion about the state of local business websites. We wanted numbers.
So on 25 September 2026 we took 100 Dundee business websites, picked at random, and put every one of them through the same checks: the basics Google looks for, the local signals that help you turn up in “near me” searches, how fast the page loads on a phone, and a few security basics. Same tests, same day, same rules for everyone.
We haven’t named anybody. This isn’t about pointing at the café with the slow website. It’s about what’s normal in this city, what’s easy to fix, and where the gap is for businesses that want to be found.
| The headline numbers | |
| Sites that passed all ten fundamentals | 1 of 100 |
| Typical score | 7 out of 10 |
| Time for the main content to appear on a phone (median) | 5.1 seconds |
| Sites taking more than 10 seconds | 32 |
| Sites telling Google they’re a local business | 18 of 100 |
| Sites that mention Dundee in their page title | 42 of 100 |
The ten fundamentals
We scored every site on ten things. None of them is advanced. Each one is either something Google has said matters, something that stops a customer getting in touch, or both. Here’s how many of the 100 passed each one.
The typical Dundee business website scores 7 out of 10. A quarter (25) score 8 or more. Another quarter (24) score 5 or less. Just 1 site passed all ten.
The good news sits at the top of that list. Almost everyone has a secure site that works on a phone. Ten years ago that wasn’t true. The problems are further down, and they’re the ones that decide whether you get found and whether anyone waits for you to load.
Speed: the biggest gap by far
We loaded every home page using the same phone and connection settings as Google’s Lighthouse test: as a mid-range Android phone on a slow 4G connection, with an empty cache. Then we timed how long it took for the main content (usually the big image or headline) to appear. That measure is called Largest Contentful Paint.
- The typical site took 5.1 seconds.
- 13 of the 98 we could time managed it in 2.5 seconds or less.
- 35 managed it in 4 seconds or less.
- 32 took more than 10 seconds, and 9 took more than 20.
The reason is almost always weight. The typical home page downloaded 1.7 MB across 67 separate files. 27 were over 3 MB, and the heaviest was about 10 MB. We re-ran the slowest three to make sure it wasn’t us. They came back within a second or two of the first run.
This test is deliberately harsh. Plenty of your customers are on good Wi-Fi with a new phone. But plenty aren’t, and a page that takes 20 seconds on a slow connection is losing people on a normal one too. For honesty’s sake: our own home page took 3.8 seconds in exactly the same test. We’re not at 2.5 either, and it’s on our list.
The fix is rarely a new website. It’s usually a handful of enormous photos, a slider nobody watches, or a pile of plugins and tracking scripts that each seemed like a good idea at the time.
Local search: the easiest wins nobody’s taking
If you’re a Dundee business, the searches that matter most are local ones: “plumber Dundee”, “restaurants near me”, “solicitor Broughty Ferry”. Here’s how many of the 100 home pages give Google (and customers) the obvious local signals.
Fewer than half mention Dundee in their page title, one of the strongest signals on the page. Fewer than one in five use the structured data that tells Google, in its own language, “this is a local business, here’s the address, here are the opening hours”. And more than half of the sites don’t let a phone user tap to call. Those are the people most likely to want to ring you right now.
Every one of these is a small job. Together they’re the difference between being a website that happens to be in Dundee and being the answer when someone in Dundee searches.
The basics that slip
- 32 sites have no meta description, so Google writes the snippet under your link for you.
- 28 have no main heading at all, even after the page has fully loaded. Another 19 have more than one.
- 18 have page titles too long to show in full in search results, and 2 have none.
- 17 don’t work properly on either the www or the non-www version of their address, so anyone who types it the “wrong” way gets an error or a different site.
- 8 don’t send visitors from the old http:// address to the secure version.
- 20 have no sitemap we could find, and 18 have no robots.txt.
- 1 site is telling Google not to show its home page in search results at all. That’s usually a setting left on from when the site was being built.
Security: the doors left open
Security headers are instructions a website sends to the browser: always use the secure connection, don’t let other sites frame this page, and so on. They take minutes to add. 32 of the 100 sites send none of the five common ones. Only 6 send all five.
20 WordPress sites announce exactly which version of WordPress they’re running, and 14 sites announce their PHP version. On its own that isn’t a hole, but it’s a signpost for anyone looking for sites running old, vulnerable software. 5 secure sites still load some files over an insecure connection.
Want to check yours? Our sister site PurpleImp.IT has a free security header check.
Being found by AI
More people now ask ChatGPT, Copilot or Google’s AI for a recommendation instead of scrolling results. We checked whether Dundee businesses were shutting those tools out. They aren’t: only 2 of the 100 block any of the main AI search crawlers in their robots.txt.
That means the question isn’t whether AI can read your site. It’s whether what it reads is clear. A site with no business schema, no location in the title and no obvious phone number gives an AI assistant very little to go on when someone asks for the best option in Dundee. The fundamentals above are the AI checklist too. (We wrote more about how much AI crawlers are reading, and how little traffic they send back, in Can I Bing it?)
Does the platform matter?
Of the 100 sites, 38 run on WordPress, 10 on Wix, 7 on Squarespace, 4 on Shopify, and 41 on something else or custom-built.
The scores barely differ between them. Wix sites scored slightly better on average, because Wix handles some basics automatically, and sites on other or custom-built platforms slightly worse. But the spread within each platform was far bigger than the difference between them. Whether anyone is looking after the site matters much more than what it’s built with.
The websites that didn’t load at all
To get to 100 working sites, we started with 130 picked at random from the Visit Dundee business directory. Of the other 30:
- 9 didn’t work at all. Six domains no longer existed, one had a certificate for a different address, one returned a server error and one returned “page not found”. Each was still the website listed for that business.
- 10 showed a “checking your browser” page to anything that wasn’t a full web browser. That’s a security setting. It usually lets Google through, but it’s worth checking it isn’t also turning away Bing and the AI tools.
- 2 refused both our crawler and an ordinary browser, and 1 timed out.
- 8 worked fine but weren’t needed once we’d reached 100.
If your business is listed anywhere with a website address, it’s worth clicking it once in a while. Nine businesses out of 130 were sending customers to a dead end.
How we did it
- The sample: businesses in the Visit Dundee directory with their own website. We left out national chains, public-sector bodies, links to Facebook or other platforms, and every site we host or look after, so our own clients can’t flatter the numbers. That left 276 eligible sites. We drew 130 at random (fixed seed) and audited the first 100 that loaded.
- What we checked: the home page only, fetched once by a clearly labelled crawler, plus the http:// and www versions of the address, robots.txt and the sitemap. None of the final 100 blocked it.
- Speed: headless Chrome, emulating a mid-range Android phone (412px wide, 4× CPU slowdown) on “slow 4G” (150 ms latency, 1.6 Mbps down), empty cache, one load per site. These are lab figures. They’re harsher than Google’s real-user data and are best read as a comparison between sites, not as a pass or fail.
- Main heading: where the raw page had no H1, we rendered it in a browser and looked again, so sites built in JavaScript weren’t marked down unfairly.
- What we didn’t do: look past the home page, check content quality, or look at Google Business Profiles. A site can score 10 and still be a poor website. The fundamentals are the floor, not the ceiling.
The anonymised data is yours to use. One row per site, every check, no names: download the CSV. If you quote it, a link back to this page is all we ask.
Want your own scorecard?
If you run a Dundee business and want to know how your site did on the same ten checks, ask us. We’ll run the test and send you the results, in plain English, with the fixes in order of what matters most. No charge and no obligation.
And if you’re one of the 25 sites that scored 8 or more: nice work. We’ll run this again next year, and we’d love to see that number go up.
Got a problem like this? Good. We like those.
Tell us what you’re dealing with and we’ll tell you honestly what we’d do.
