How to tell whether your website was built badly

Nearly every rebuild conversation I have opens the same way. Someone shares their screen, scrolls their homepage, and says some version of "it looks a bit 2018, doesn't it".

Sometimes it does. It's almost never the thing costing them money.

The worst-built site I've looked at this year was handsome. Nice photography, restrained type, someone had clearly cared about it. It also hadn't delivered a single enquiry email since a hosting change the previous spring, and the owner had spent that whole period assuming the quiet was seasonal. He found out because a customer rang him, annoyed, to ask why nobody had got back to her.

So here's what I'd actually check, in the order I check it. None of it needs a developer and you can get through the lot in about twenty minutes.

Dated and broken are different problems

Worth separating these before you start, because they get treated as one thing and they have completely different price tags.

A dated site is a design problem. Old fonts, a stock photo of a handshake, a layout that was fashionable when it was made. It's fixable, it's fairly cheap to fix, and it might not even be urgent depending on who your customers are and what your competitors look like.

A badly built site is a construction problem. It can look immaculate. The rot is in the things you can't see from the front: whether the form works, whether anyone but the original developer can change a word of it, whether it's carrying a hundred kilograms of code it doesn't use, whether Google can read it. You can spend a fortune restyling a site like that and change nothing that matters.

The reverse happens too, and it's the happier outcome. Plenty of ugly sites are structurally sound and just need a designer, which is a much shorter job than a rebuild.

Start with the form, because it fails silently

Go and send yourself an enquiry. Right now, before you read the rest of this. Use your phone, put a real message in it, and then wait ten minutes and check whether anything arrived.

I put this first because form failure is common, it's invisible, and every day it continues costs you actual work. The site keeps loading. The thank-you message still appears after you hit send. From the visitor's side there is no sign at all that their enquiry evaporated.

The usual causes are dull. A notification address belonging to someone who left. A sending service whose free tier lapsed. A domain move where the email side was never reconnected, so messages now land in spam or get rejected outright. A third-party form widget that was discontinued.

If nothing arrives, that's your whole project for this week and everything else in this article can wait. And when it's fixed, make sure the form sends a plain copy to a shared address as well as doing anything clever, which is the argument I made in what to connect first. The clever version is where things break; the dumb copy is how you find out.

Load it on your phone, on mobile data

Turn the wifi off first. This catches more owners than you'd think, because you and your developer have always looked at this site over a good connection in the same city as the server, and your customers haven't.

Then run it through PageSpeed Insights, which is free and takes a minute. The number I'd look at is Largest Contentful Paint, which is roughly how long until the main thing on the page turns up. Google's guidance is that LCP should be 2.5 seconds or less for a good experience, and if a site is coming back at three or four times that on mobile you have found something real.

Slowness is one of the few problems where you don't need to interpret anything. It's costing you visitors who left before they read a word, and we went through what that's worth in why your website is slow.

What I'd note is where the weight comes from, because it changes the diagnosis. Enormous unoptimised photos are a bad afternoon's work to fix and nothing more. Slowness that comes from how the site was assembled, where every page drags in scripts for features that page doesn't use, is a build decision, and you can't tune your way out of it.

Try to change one sentence

Pick a sentence on any page. A phone number, a line of the about page, whatever. Now try to change it yourself.

This is the test that predicts the next five years, and it's the one nobody runs. If you can find it, edit it, and see it live, then whatever else is wrong with the site, you own it in the way that counts. If changing a sentence means emailing someone and waiting, then every small correction has a cost and a delay attached, and what actually happens is that you stop making corrections. The prices go stale. The staff list has someone on it who left in 2023. The site slowly stops being true, not because anyone decided that, but because the friction was too high.

I'd rather inherit a plain site the owner can edit than a beautiful one they can't touch. It isn't close.

While you're in there, work out who holds the logins for the site, the domain and the hosting, because a site nobody can get into is a different kind of badly built and it's the one that gets urgent without warning.

Then check that Google can see it

Search site:yourdomain.com.au in Google. You'll get a rough list of what's actually indexed.

If your main service pages aren't in there, that's a real finding, and it's a separate topic with its own set of causes, most of which are boring and fixable. The triage is in why your website isn't showing up on Google. What matters for this exercise is only whether the pages exist as far as Google is concerned, because a page that isn't indexed can't do anything for you no matter how well it's written.

Repair or replace

Rough rule from doing a lot of these: forms, speed from oversized images, content that needs updating and the design itself are all repairs. Sometimes tedious repairs, but repairs.

What tends to push a site over the line is structural. You can't add a page without paying someone. The content only exists inside a template nobody can decipher. It's slow because of what it is rather than what's on it. Nobody can get into the accounts. Any one of those on its own might be survivable. Two or three together and you're paying to maintain a problem instead of fixing it, and rebuilding usually works out cheaper over any sensible timeframe.

If you'd rather someone else ran the twenty minutes, that's most of what a free website audit is, and we'll tell you plainly which column you're in.

One last thing, and I mean it. The person who built this almost certainly wasn't incompetent. Sites get built to a budget, in a hurry, by someone who was never engaged to look after it afterwards, and then five years happen. Working out what you've got is bookkeeping, not a verdict on anyone.