There are plenty of ways to get a website.

You can choose a template.

You can drag blocks around until something looks about right.

You can describe what you want to AI and wait for it to produce something remarkably similar to every other website it has ever seen.

All perfectly legitimate ways to get online.

They’re just not how I work.

I start with a blank page. Not because I enjoy making life difficult for myself, but because I don’t want the website deciding what the business should look like before we’ve even started.

The business comes first

Before thinking about layouts, buttons or clever bits, there’s a much more useful question:

What does this website actually need to do?

Who is it for?

What do they need to find?

What does the business need them to understand?

What needs to stand out?

And what doesn’t?

Those answers shape the website. Not the other way around.

That’s the difference between designing a website and filling one in.

Nothing has to be “close enough”

One of the great things about starting from scratch is that you don’t have to compromise because a template has decided that something belongs in a particular place.

If the navigation needs to work differently, it can.

If the content needs a different structure, it can.

If a page needs one large thing rather than three little things because three is what the template happens to offer, that’s fine too.

There is no template waiting patiently for you to stop messing with it.

It’s also worth asking what “a thousand templates” actually gets you. Somehow the local dentist, the local accountant and the café down the road all end up choosing the same one. A thousand options, and everyone lands on the same three.

The design isn’t a skin

I don’t see design as something you put on top of a website once all the technical stuff has been sorted out. The two are connected from the start. How something looks affects how it works, and how it works affects how it needs to be designed.

A layout that looks great on a desktop might be rubbish on a phone. A beautifully balanced mock-up might fall apart when the real heading is twice as long. A clever animation might be completely pointless once you actually use the page. These are the kinds of problems that only show up once you’re actually building the thing, not while you’re admiring it in a static mock-up.

Anyone who’s used a drag-and-drop builder knows the horror of nudging one image two pixels to the left and watching the entire mobile layout quietly fall apart. Peek under the hood of one of these sites and you’ll usually find a maze of nested divs, built by a system trying to be generic enough for everyone, which is a different thing entirely from being built for you.

Which is why I do both. It’s much easier to solve a problem when the person who designed it also has to make it work, and the code underneath only has to make sense for one website.

The content gets a say

Real content is inconvenient. It refuses to arrive in neat little boxes of exactly 42 words.

Some businesses have three services. Some have thirty.

Some have one photograph worth using. Others have hundreds.

Some need a simple contact form. Others need booking, product information, calculators, integrations, or something nobody has thought to give a name to yet.

A bespoke website can accommodate that. The content doesn’t have to apologise for not fitting the template.

The same goes for how the content is written. A lot of platforms hand you an SEO checklist and treat it like a recipe: keyword in the title, keyword in the first paragraph, keyword density somewhere between suspicious and unreadable. Tick every box and the plugin turns green, whether or not an actual human would enjoy reading the result. I’d rather write for the person landing on the page and trust that Google is clever enough to notice.

You don’t need everything

Most website platforms have to cater for everyone, which means they give you everything. Menus for menus. Settings for settings. A dozen ways to build a gallery. Twenty-seven options for how a button behaves.

I’d rather build what the website actually needs. If something isn’t needed, it doesn’t have to be there, which keeps things simpler for everyone, including the person who has to maintain the website later.

It also tends to keep things faster. Stack enough plugins and features nobody’s using onto a homepage and the load time can stretch long enough that a visitor could have driven to your shop and back. And the more moving parts there are, the more there are to break. Sometimes all it takes is one automatic plugin update, applied while nobody’s looking, to bring the whole site down. A bespoke site doesn’t have a tower of plugins to begin with, so there’s nothing there to wobble.

Your website shouldn’t require a manual

A content management system should make updating a website easier. It shouldn’t feel like you’ve been handed the controls to an aircraft.

With a bespoke site, the editing system can be designed around the content.

If you need to add a staff member, there’s somewhere to add a staff member.

If you need to update a service, you update the service.

You don’t need to understand how the page was built, or worry that changing a heading will somehow move the footer to Argentina. The complicated stuff just stays underneath.

There’s no discovering halfway through, either, that something as basic as a working contact form is sitting behind a premium tier you haven’t paid for. If you need people to be able to email you, that’s not an upgrade. That’s a website.

If it needs to do something, we can work it out

This is probably my favourite part.

With a template or website builder, you often start by asking:

“Can it do that?”

If the answer is no, you start looking for an extension. Then another extension. Then something to make the two extensions talk to each other.

With a bespoke website, I can start somewhere much more useful:

“What do you need it to do?”

Then we work out how to build it. That’s particularly useful when a business has requirements that don’t fit neatly into a standard website.

It doesn’t have to look like a website

This is where design really matters.

There are a lot of websites that are perfectly competent. They work. They look professional. They have the obligatory hero image, centred white text and a single button sitting in the middle of it all, the unmistakable look of a website built in an afternoon. Nothing wrong with any of that. It’s just been done. A lot.

A website should communicate something about the business behind it. It can have personality. It can be quiet. Clever. Direct. Serious. Odd. It can simply look different, not for the sake of being different, but because there was never a reason for it to look like everybody else.

AI can build you a website. Keeping it is another matter.

AI is very good at patterns. Give it enough examples and it’ll produce something that looks convincing. But if the instruction is “make it modern, professional and trustworthy”, you’re probably going to get something that looks modern, professional and trustworthy. You might also get something that looks remarkably familiar to the last one it built, and the one after that.

That’s the design side. The more interesting problem shows up once the site is actually live.

Most AI website builders don’t hand you a website. They hand you a subscription. The site lives on their servers, built with their editor, and if you want to change so much as a heading, you go back into their system and ask their AI to do it for you. There’s no file to open, no code to hand to someone else if you outgrow them. Just another prompt, and another month’s invoice.

Want to leave? Good luck. Plenty of these platforms won’t let you export your design or code at all, so “moving your website” means rebuilding it somewhere else from scratch. And the pricing has a habit of shifting once you’re settled in: cheap to get started, considerably less cheap once you’ve built something you’d rather not lose.

None of that makes AI useless. It’s a genuinely handy tool for plenty of things. But “build a website with AI” often quietly means “rent a website from AI”, and the rent doesn’t stop.

The interesting bit, deciding what the business needs to say, how it should say it, and what people should feel when they arrive, is still the part I like doing. It’s also the part a pattern-matcher can’t really do for you.

So why do I build websites this way?

Because this is the work I want to do.

I’m not interested in selling hours of dragging boxes around a screen. I’m interested in solving the problem before opening the toolbox.

What should the website be? How should it work? What should it say? What should it feel like? And what will the client need to change six months from now?

Then I design it, and I build it.

It’s more involved than choosing a template. It takes more thought. And it costs more than doing it yourself.

But if your website is an important part of your business, I think it’s worth treating it like one.

Not as something you fill in. Something you make.