Great brand. Broken build.

A friend sends you a link to a restaurant. Everyone's been talking about it. The photos on Instagram look incredible, the interiors are the kind you screenshot for later. You tap the link on your phone, standing in a queue, and wait. The screen is white. Then the logo appears. Then, eventually, a menu. By then you've already put the phone away and suggested somewhere else. The restaurant never knew you tried. There's no record of the table that didn't get booked.

That small, ordinary moment is the whole subject of this article. We wanted to know how often it happens, so we went and measured it. Not on struggling businesses with a website a cousin built in 2014. On the names Cape Town knows, alongside plenty it doesn't.

01Twelve kinds of business, one test

We picked 52 Cape Town business websites across twelve sectors: restaurants and cafés, wine estates, dentists, plumbers and electricians, yoga and pilates studios, guesthouses, designers, retailers, tax and accounting firms, property, conservation charities, and a helicopter company. Twenty-three of them are what we'd call well-known, the kind of name you'd recommend to a visitor without looking it up.

Every site got the same treatment. We ran it through the same audit engine that powers the Digital Health Score on our own site, then through Google's Lighthouse test, set up the way Google's own mobile tests are: a modest phone on a slow 4G connection. One run per site. Nobody was singled out, and nobody is named here, because that was never the point. The point is the pattern.

02The wait

Lighthouse scores speed out of 100, and its "good" band starts at 90. Out of 52 sites, six got there. Nearly nine in ten were slower than they should be. Among the 23 well-known names, exactly one passed. The middle score for that group was 56.

The businesses you'd expect to care most about a first impression did no better. The hospitality names in our sample, the places that live and die on someone fancying a table or a tasting, had a middle score of about 55. A third of them scored under 50. These are businesses that will spend a fortune on a plate and a candle and then lose the booking before the page has finished loading.

Scores are abstract, so here is the same thing in seconds. Google asks for the main content of a page to show up within about two and a half seconds. On our test, 43 of the 52 sites took more than four. Twenty-three took more than ten. The slowest took nearly half a minute, 29.4 seconds, to show a visitor the thing they came for.

A fair caveat before anyone gets defensive: that test is deliberately harsh. A modest phone on a slow connection is a stress test, not a stopwatch, and your real customers on fibre will see something kinder. But plenty of real customers are not on fibre. They are on a phone, on mobile data, in a car park, deciding in about three seconds whether you're worth the effort.

03Two budgets, two different people

Here's what we think is going on, and it's an opinion, not a finding. Most businesses that care about how they look spend real money on it. A photographer. A designer. A brand identity with a story and a typeface and a mood board. That budget is easy to justify because you can see what it bought.

The build is the other budget, and it's invisible when it's done well. Nobody frames the compression settings. Nobody posts a screenshot of a fast load. So it quietly loses the argument every time, and the heavy photographs the brand budget paid for often get shipped to a phone at full size.

The brand budget bought the photographs. Nobody bought the loading. The person who approved the site judged it on a large screen, on office wifi, with the files already cached. The person who actually has to use it is a different person in a different place, and they never get a say in the sign-off.

04The boring things they got right

It would be easy to read all this as "big names are careless." The data doesn't say that. On the technical basics that sit behind the scenes, the well-known businesses did better than everyone else.

Take HSTS, a security setting that tells browsers to only ever connect to your site securely. About half of the well-known names were missing it. That sounds bad until you see that nearly eight in ten of the other sites were missing it too. Someone competent set up the hosting, the certificates and the plumbing at the better-known businesses. The thing nobody got to was the weight of what they were serving on top.

That distinction matters. It means the problem isn't carelessness or small budgets. It's that nobody owns the question "how does this feel on a phone?" Design owns how it looks. Hosting owns whether it's up. The gap in between belongs to no one.

05Pretty doesn't predict anything

We expected the design studios and retailers to do best. They had the most to prove. Three of the six sites that passed did come from that group, which looks like confirmation until you look at the spread. That group of eight ranged from a 35 to a perfect 100. Same trade, same eye for detail, opposite result.

It's a small group, so we wouldn't build a theory on it. But it fits what we see in client work all the time: visual quality and technical quality are separate skills, and having one tells you very little about the other.

The same applies to people who can't see the screen at all. Thirty-four of the 52 sites had most of their images missing alt text, the short description that lets a screen reader tell a blind visitor what a photo shows, and that Google reads to understand the page. Two in three. For a restaurant whose entire pitch is how the food looks, that's a lot of invisible dinners.

06What this isn't

This is not a scientific survey. We chose 52 sites by hand, we ran each test once, and speed on any given day moves around. Deciding which businesses count as well-known was our judgement, and we're not publishing the list. Lighthouse is a simulation, not a measurement of any one visitor's experience.

One more honest note. When we compared the well-known names against everyone else, the speed gap between the two groups was too small to call real. So we're not saying famous businesses have slower sites than unknown ones. We're saying something simpler and, to us, more interesting: being well-known didn't help.

07What a better build looks like

None of this is exotic. Images sized for the screen they're actually shown on, not for a billboard. Fonts and scripts that load only what the page needs. A hosting setup chosen on purpose. And someone, ideally the same someone who designed the thing, checking how it behaves on a mid-range phone before it goes live and again every few months after.

If you're commissioning a site, ask one question before you sign off the design: what will this score on a phone, and who is responsible for keeping it there? A good agency will have an answer. A great one will already have tested it. If the room goes quiet, you've learned something about who is going to own the gap once the invoice is paid.

The test for your own site takes about a minute. Open it on your phone, on mobile data, with Wi-Fi switched off. Count. If you'd have given up before it finished, so did someone who was about to become a customer.

If you'd like us to look at what's actually happening on yours, that's what we do. Start a project and we'll tell you honestly whether it's the build, the brand, or both.