Should you build your own website?

The honest case for doing it yourself

Worth saying first, because it's true and it usually goes unsaid by people in my trade: for some businesses, building it yourself is the correct decision.

• You are pre-revenue or testing an idea. Spending thousands to validate something you might abandon in four months is a bad trade. Build the cheap version.
• Your site is genuinely a business card. Name, what you do, phone number, hours. If nobody finds you through search and never will, the invisible layer this article is about matters much less.
• You will actually maintain it. An owner who updates their own site weekly beats a beautiful agency build that went stale in 2023. This is more common than anyone admits.
• You enjoy it. Not a small thing. People do good work on things they like doing.

The modern builders are also genuinely good. Squarespace, Wix, Framer and the rest produce sites that look far better than most agency work did a decade ago. If the objection to DIY were “it will look amateur”, that objection expired years ago.

“I could do it, I just don't have time”

This is the sentence I hear most, and it's true — you almost certainly could, and you almost certainly don't have time. But I think it's the wrong reason, and it quietly leads people to the wrong decision.

Because if time is the only obstacle, the logical move is to find a weekend. And plenty of owners do exactly that, produce something that looks good, and then quietly conclude a year later that websites don't work for their business. The site wasn't the problem. The weekend wasn't the problem either.

The problem is that you finished. The editor showed you a complete, attractive page, and nothing anywhere told you that a substantial part of the job hadn't been started — because the tools that would tell you're not in the builder, and nobody selling you the builder has any reason to mention them.

The half you can't see

Roughly half of what determines whether a site performs is invisible in an editor. Not hidden — genuinely not represented anywhere in the interface you built the site in. Here's what I most often find, in rough order of how much damage it does:

• Images shipped at full camera resolution. A 4MB photo scaled down by the layout still downloads as 4MB. The page looks identical and takes eight seconds on a phone.
• No structured data. Nothing telling a search engine what your business is, where it is, what you sell or who wrote the page. It can't be inferred reliably and most builders don't add it.
• Titles and descriptions left on the default. Often the page name, or the same sentence on all fourteen pages. This is the text people read in search results and decide whether to click.
• Pages nothing links to. Built, published, unreachable from any menu. Search engines struggle to find them and so do people.
• Mobile behavior that was never tested on mobile. Menus that don't open, tap targets too close together, forms that won't submit on a phone. Desktop preview at a narrow width isn't the same thing as a phone.
• Layout that jumps while it loads, because images and embeds have no declared dimensions. You won't notice on your own fast connection with everything cached.
• Invalid markup. Browsers are forgiving, so it looks fine. Screen readers and search engines are less forgiving.

Every one of these is invisible from inside the editor. That's the entire point — not that they're difficult.

This isn't really a DIY problem

Here's the part that surprises people, and the reason I don't think any of the above reflects badly on business owners who build their own sites.

On 20 August 2026 I ran Google's PageSpeed Insights against six professional web design and SEO firms serving this region, all on the same mobile test. Their scores ranged from 28 to 80. One shipped a home page of over 10MB. One had a layout shift score five times worse than Google's “poor” threshold — on a phone, roughly half the screen moving while you try to read it. Several took more than eight seconds to show their main content.

These are firms that sell websites for a living, on their own sites, where they control everything and are presumably showing their best work.

So when an owner-built site has these problems, it's not evidence that the owner was out of their depth. It's evidence that this layer gets skipped by almost everybody, because it's invisible, unglamorous, and nobody notices its absence until they wonder why the phone isn't ringing.

The difference isn't that professionals always get it right. It's that they should know it exists. The measurement that produced those numbers is explained here, and you can run it yourself in a minute.

Eight checks you can run this afternoon

Rather than take my word for any of it, here's the audit. It takes about twenty minutes, needs no technical knowledge, and it's the same set of things I look at first.

1. Measure it. Run your address through the speed test on this site or pagespeed.web.dev. Pass: above 50, and the main content appearing in under 4 seconds.
2. Use it on a real phone, on mobile data. Turn wifi off first — this is the single most revealing test on the list. Open the menu. Tap every link in it. Submit your own contact form. Pass: everything works and nothing needs pinching to hit.
3. Search site:yourdomain.com in Google. That lists the pages Google has actually indexed. Pass: the number roughly matches how many pages you think you have, and the important ones are there.
4. Read your own search results. In that same list, look at the blue titles and gray descriptions. Pass: each one is distinct, readable, and describes that page rather than your business in general.
5. Test your structured data. Paste your home page into Google's Rich Results Test. Pass: it finds something. Most owner-built sites return nothing at all, which is the single most common finding in this list.
6. Validate the markup. Run it through the W3C validator. Pass: not zero errors — almost nothing has zero — but not hundreds, and no unclosed tags or duplicate IDs.
7. Navigate with the keyboard. Press Tab repeatedly from the top of the page. Pass: you can see where you're at every step, and you can reach the menu and the form without a mouse.
8. Watch someone else use it. Ask a person who has never seen it to find your pricing, or book, or contact you. Say nothing while they try. Pass: you do not have to intervene.

What the result actually means

If you passed most of them: your site is in better shape than a majority of professionally built ones, including several I measured for this article. Don't let anybody — me included — talk you into a rebuild. Fix the one or two that failed and get on with your business.

If you failed a few: almost all of these are individually fixable without rebuilding anything. Compressing images, writing real page titles, adding structured data and fixing a mobile menu are afternoon jobs, not projects. Whether you do them or pay somebody is a question about your time, and there the “I don't have time” answer is a perfectly good one.

If you failed most of them: that's worth knowing, and it's still not an indictment of your work. It means the site was built with tools that never surfaced this layer, which is the normal outcome and not a personal one. The question then is only whether the site is meant to bring you work. If it's a business card, leave it. If you need it to earn, this is the gap between the two.

The reason I'd rather you ran the checks than took my word for it's straightforward: I sell this work, so my opinion about whether you need it's not worth much. The measurements don't care who's reading them.