Six things wrong with your website that have nothing to do with the design

I have a habit I cannot switch off. Someone sends me their site to quote on, and before I have looked at a single design decision, I have already run the same six checks. They are the same six every time, and almost never all clean. Not one of them shows up in a screenshot, which is precisely why they survive, and why so many businesses end up buying a redesign that fixes none of them.

Most websites that underperform are not failing on design. They fail on six things a screenshot cannot show: nobody can find the site, the contact form asks too much, someone else owns the domain, a rebuild was bought before a diagnosis, the site looks abandoned, and a share of visitors physically cannot use it. Fix those and you are ahead of most of your competitors without touching a font.

Can anyone actually find your site?#

Go to Google right now. Type the thing you sell plus your town. Durbanville plumber. Fixed asset software South Africa. Look at page one.

If you are not on it, you are not in the running, because almost nobody clicks through to page two. And if your buyer asks ChatGPT instead, the same problem gets worse: they get three recommended businesses, and if you are not among them, there is no page two to be missed on at all.

What annoys me is how rarely this gets raised before the invoice. Designing a site and making it rank are separate disciplines, and the second one is frequently not priced into the job. So a business spends R40,000, launches, waits for the phone, and ends up with an expensive pamphlet that only existing customers ever see.

I wrote up the exact method I use to get a page cited by AI search if you want the long version. The short version: if search was not part of the brief, assume it was not part of the build.

Is your contact form costing you leads?#

Someone has decided they want to talk to you. They are sold. They click the button.

And then you show them fourteen boxes.

Name, email, phone, company, role, budget range, project timeline, how did you hear about us, preferred contact method, a captcha that fails twice. Every field you add costs you people. That is one of the most tested findings in web design, and the often-quoted example is Expedia removing a single field from a checkout and reportedly recovering millions a year in revenue.

On a first contact you need a name, an email, and one line about what the person wants. That is it. You can ask everything else on the call, where the same questions feel like interest rather than a toll gate.

Then do the bit almost nobody does: fill in your own form and check that it lands in the inbox you actually read. Form delivery breaks quietly and often. Someone leaves and their mailbox is deleted. The domain moves host and the mail records do not follow. None of that generates an error. There is no bounce, no warning, just enquiries going nowhere until somebody notices months later.

Do you actually own your own website?#

This one makes people go quiet.

Three questions:

1. Is the domain registered in an account you can log into right now?
2. Do you have access to the hosting and the site itself, not just a page editor someone else set up?
3. Could you export the whole thing and move it somewhere else tomorrow?

If the honest answer to any of those is “I would need to ask the person who built it”, you do not own your website. You are renting it from someone who has no contractual obligation to keep answering the phone.

Hosting you can rebuild in a day. Content you can rewrite. The domain is the piece you cannot simply redo, because it carries your email, your Google listing, and every link anyone has ever built to you, and prising it out of a third party’s registrar account goes through a dispute process, not a conversation.

This is why everything I build lives in the client’s own accounts, on their domain, under their billing. Not as a favour. It is the only arrangement that is not a hostage situation waiting to happen.

Do you need a rebuild, or a diagnosis?#

Business slows down. The “we need a new website” instinct kicks in. It is the most expensive way I know to fix a problem nobody has diagnosed.

Here is the pattern. R80,000 and three months go into a beautiful rebuild. It launches. The numbers do not move, because the thing holding the business back was the form from two sections ago, or a homepage that never said plainly what the company sells. Now the budget is gone, the diagnosis still has not happened, and there is nothing left to spend on the actual problem.

I say this as someone who sells rebuilds. Most of the time you should fix the leak first. It costs a fraction, it takes days instead of a quarter, and it tells you what is really wrong. A rebuild earns its keep when the foundation genuinely cannot carry the business any more: the platform is unbearably slow, you cannot edit it without a developer, the structure fights every new page you add. Not because the design feels a bit 2019.

Order matters. Diagnose, fix the cheap things, measure, then decide. And if the diagnosis does point at the build, you at least know what the rebuild has to solve: a site that loads in under two seconds rather than a prettier version of the same problem.

Does your site look like you have closed down?#

There is a version of your website that quietly tells strangers you went out of business, and you will never see it, because you already know the business is alive.

The loudest signal sits in the browser bar. No security certificate means every visitor gets the words not secure next to your address, right as they are about to type an email or a card number. Nobody types a card number into a page that is warning them about itself. Certificates are free and take minutes.

Then the rest of it: a copyright line stuck on 2022, a “latest news” box whose newest item is three years old, two or three links that go nowhere. To you, invisible. To a stranger deciding whether to trust you with money, it reads as a business that quietly wound down two years ago.

None of this costs money. It costs an afternoon. Get the certificate. Make the footer year generate itself instead of being typed. Run a free broken link checker. And if there is a blog you are never realistically going to write again, take it down, because an abandoned one does more damage than no blog at all.

Can everyone actually use your site?#

About one in six people worldwide lives with some form of disability, according to the WHO. Eyesight, colour vision, motor control, or needing a screen reader.

If your images have no alt text, the screen reader has nothing to say. If your menu only opens on hover, keyboard users cannot get past it. Those visitors do not email you to complain. They leave, and they show up in your analytics as a bounce, indistinguishable from someone who was never interested.

The legal side is real too, though it lands differently depending on where you sell. The European Accessibility Act has been enforceable since June 2025 with fines behind it, and US businesses face thousands of ADA web lawsuits a year, plenty of them small firms. South Africa has no equivalent enforcement regime for private-sector websites yet, so here the argument is simply commercial: more people who can buy from you.

Three things get you most of the way. Label your images. Make sure the whole site works with a keyboard and no mouse. Check your text contrast. Then run a free accessibility checker; it will find things neither of us would think to look for. I hold Zivaro’s own work to WCAG 2.2 AA, and the honest surprise was how little extra time it costs when it is in the build from day one rather than retrofitted afterwards.

Why does none of this ever get fixed?#

Look at the six together and the pattern is obvious. Findability, the form, the domain, the diagnosis, the padlock, the keyboard. Not one of them is a design problem. Not one shows up in a screenshot.

That is precisely why they survive. Owners assume the person who built the site would have mentioned anything important. Builders assume anything important would have been in the brief. Nobody raises it, and the six sit there costing money and credibility, year after year.

It is a strange thing to do to an asset that works every hour of the week. You sign the design off once, leave it unsupervised for five years, and the only time anyone looks again is to ask whether it looks dated.

So look at the other six instead. They cost nothing to check, none of them needs a designer, and finding three that fail is three more than a rebuild would ever have told you. If you would rather I ran them, send me the URL. I will tell you what I find, including when the answer is that you should not spend the money yet. It often is.