A week ago, we were chatting to a plumber who was considering us to build his new plumbing website. During that conversation, he asked a perfectly reasonable question. “Why don't you build your websites using AI? Some of the other companies I've looked at do.”
The answer that came to mind was fairly simple. We asked him whether he would tighten a nut with his fingers. He laughed. Of course he wouldn't. He might get it started, but if he wanted the job done properly, he would use the right tool.
That is broadly how we look at AI generated websites. AI is becoming increasingly common in web design, including among companies building websites specifically for tradespeople. It can generate layouts, structures and pages of copy remarkably quickly. And we could use it to build websites that way too. It would make our job considerably easier. We could produce websites faster, reduce the amount of design and development work involved and move on to the next project sooner.
But that is not how we want to build websites. The foundations of every Toolbelt website are designed and built by hand, using more than 40 years of combined web design and development experience, and we pride ourselves on that.
We decide how the website should be structured. We decide what information deserves prominence. We plan how customers should move through it. We build the functionality. We consider how it will work on different devices and how it can change as the business grows.
Most importantly, we start with the business. Its services. Its customers. Its locations. Its priorities. The type of work it wants more of and what its customers need to know before getting in touch. That approach sits at the heart of our website design for plumbers and every other trade website we build.
We build the website around the business.
We don't build the business around whatever a piece of software generated.
AI can build a website. That isn't really the question.
There is no point pretending AI cannot build websites. It can. Give an AI website builder a business name, a trade and a few instructions and it can produce something that looks like a website remarkably quickly.
It can suggest a structure. It can create sections. It can write headings and paragraphs. It can decide where buttons should go and produce something that, at first glance, may look perfectly respectable. That technology will continue to improve. But we think there is a much more important question.
Is generating a website the same thing as properly designing and developing one?
We don't think it is. When an experienced web designer looks at a business, there are dozens of decisions being made that are difficult to reduce to a prompt.
- Which service should be most prominent?
- Should the main action be a phone call or a quote request?
- Does an emergency plumber need the same customer journey as a bathroom installer?
- How much information should be visible before somebody is asked to contact the business?
- Which services deserve their own pages?
- What should happen on mobile?
- What happens if the business adds another service next year?
- What if the owner wants to change the layout, add a calculator, expand into another area or alter the way customers enquire?
Those are not simply design questions. They are business decisions expressed through a website.
A lot of trade websites are now being generated with AI
This is one of the reasons we have taken such a clear position on it. Before AI website builders became commonplace, we were already seeing the consequences of websites that had been built too quickly or without the right foundations. We spent years helping tradespeople repair WordPress websites where corners had been cut during the original build. Sometimes the website looked perfectly acceptable when it first went live, but underneath it there were problems.
- Plugins conflicted.
- Contact forms stopped working.
- Layouts became difficult to change.
- Updates caused things to break.
- Important functionality had been bolted on rather than properly planned.
In some cases, the customer had effectively been left with a half finished product that was difficult to maintain and even harder to develop as the business changed. We have seen websites start developing serious problems within a year or 18 months of launch, not because WordPress itself was the problem, but because of the way the website had originally been put together.
That experience has heavily influenced the way we approach website design for tradesmen today. The technology has changed, but in some cases we are now seeing a very similar problem appearing in a different form.
We are increasingly speaking to businesses that have had a new website built relatively recently, sometimes within the last 12 months, and are already looking to replace it. Some of those websites have clearly been produced using heavily automated or AI driven processes. And importantly, these are not always websites that were sold cheaply.
The customer may have paid a perfectly respectable amount of money for the finished product, only to later discover that the website is difficult to update or change, important functionality does not work reliably, or nobody seems able to make relatively straightforward updates. Shockingly, in some cases we have even found the wrong contact number in the customer's footer. We have seen contact forms that were not delivering enquiries properly.
We have spoken to customers who could not make the changes they needed. And we have seen businesses reach the point where replacing a relatively new website makes more sense than continuing to work around its limitations. That is exactly the situation we want to avoid.
What can happen when the foundations are weak
Broken form
Difficult update
Layout issue→Rebuild
Built with future changes in mind
AI generated websites are becoming increasingly common in the trades website market. Some companies are using AI heavily to generate the initial design, page structure and written content. From a production point of view, it makes complete sense. It is fast. And when you are producing websites at scale, speed matters.
The problem is that when similar systems are given similar instructions about similar businesses, the results can start to feel remarkably familiar.
A plumber needs a homepage.
Add three service boxes.
Add a section about being trusted and reliable.
Add some reviews.
Add an About section.
Finish with a big call to action.
Change the logo, colours and company name.
Next.
There is nothing inherently wrong with that approach. For some businesses, it may be perfectly adequate. It just isn't what we're selling. Toolbelt Websites was not created to produce as many websites as possible in the shortest amount of time. We created it to build good websites for tradespeople.
The copy is often where AI becomes obvious
One of the easiest places to spot an AI generated trade website is often the writing. You have probably seen phrases like these hundreds of times:
Your trusted partner for all your plumbing needs.
Or:
We pride ourselves on delivering high quality workmanship and exceptional customer service.
Or:
From small repairs to complete installations, no job is too big or too small.
There is nothing grammatically wrong with any of them. That is almost the problem. They could belong to practically anybody.
Change the company name and the same paragraph could work for another plumber, an electrician, a roofer, a landscaper or a builder. It tells you very little about the actual business.
Try the name swap test
Generic copy
“Trusted, reliable and professional.”
Business-specific content
- Actual services
- Actual locations
- Actual process
- Actual customer questions
There is a simple test we like. Take a paragraph from your website and replace your company name with one of your competitors. If the paragraph still works perfectly, ask yourself whether it is really saying anything distinctive about your business. Good website content should come from the business itself.
- What work do you specialise in?
- What sort of customers do you work with?
- Which areas do you genuinely cover?
- What type of enquiries do you want more of?
- What happens after somebody contacts you?
- What do customers regularly ask before booking?
- What makes dealing with your business different?
Those answers give us something meaningful to work with. “Trusted, reliable and professional” doesn't.
We don't start with a template for what a tradesman should look like
There are conventions in web design for good reason. Visitors expect navigation to behave in a familiar way. They expect contact information to be easy to find. They need pages to work properly on a phone. We are not trying to reinvent websites for the sake of being different. But there is a difference between using proven design principles and forcing every business into essentially the same website.
We start by understanding the individual business before deciding what the website needs. A self employed plumber who relies heavily on emergency callouts does not necessarily need the same website as a plumbing and heating company selling large boiler installations. An electrician targeting domestic rewires has different priorities from one looking for commercial maintenance contracts. A landscaper selling £20,000 garden transformations needs to build confidence differently from somebody offering regular garden maintenance.
The trade might tell us where to start. It shouldn't dictate where we finish.
Of course, there are elements of a website where a proven, repeatable approach makes sense. Navigation needs to be intuitive. Footers need to contain useful information. Contact pages need to make getting in touch straightforward. We are not trying to reinvent the wheel. We are building websites to serve a purpose and to give tradespeople the best possible chance of turning visitors into genuine enquiries.
Experience still matters
Between us, we have more than 40 years of combined experience in web design and development. That experience matters. Not because everything was better before AI existed. It wasn't. Technology has made web development faster, more capable and more accessible, and that is a good thing.
But experience teaches you where problems tend to appear. It teaches you that a design can look excellent on a laptop and be frustrating on a phone. It teaches you that something seemingly simple today can become difficult to maintain once a business wants to change it.
It teaches you when a clever visual effect is actually getting in the customer's way. It teaches you when a page needs more information and when it needs less. And perhaps most importantly, it teaches you that two businesses in the same industry can need very different websites.
AI can generate options extraordinarily quickly. Experience helps you decide which option actually makes sense.
We want websites that can change with the business
A new website represents the business as it exists today. But businesses don't stand still. A plumber might initially offer general plumbing and boiler servicing. Six months later they may want to promote bathroom installations.
Later they might take on another engineer, cover a wider area or introduce heat pumps. They may want to add project galleries, change how quotations work, add new service pages or completely alter the emphasis of the homepage.
The website needs to be able to accommodate that. That is one of the reasons flexibility matters so much to us when we build. We don't like the idea of giving a customer something that works only as long as they leave it exactly as it was on launch day. Nothing on the web is truly future proof.
Technology changes.
Search engines change.
Businesses change.
Customers change.
But a website can absolutely be built with future changes in mind. That is what we aim for.
Faster is not automatically better
One of the strongest arguments for AI generated websites is speed. And it is a legitimate advantage. If a system can generate most of a website in a few minutes, the company producing it can deliver more websites with considerably less manual work. We understand why that model exists.
But there is an important distinction between something being faster for the supplier and something being better for the customer. If we automated more of our production process, it would benefit us.
- We could build more websites.
- We could reduce the amount of time we spend making design decisions.
- We could generate large amounts of content rather than writing around the individual business.
- We could move projects through production much faster.
But maximising the number of websites we can produce isn't our objective. We would rather take the time to build something around the individual customer. That might not be the fastest way to run a website business. We think it is a better way to build websites.
What about templates?
Hand built does not mean ignoring everything we have learned from previous websites. Quite the opposite. Over time, good web designers develop systems.
We learn which navigation patterns work well. We know how to build responsive layouts properly. We know how forms should behave. We know how to structure common components and how to make websites manageable after launch.
We reuse good engineering. What we don't want to reuse blindly is the customer's website. There is an enormous difference between having a solid technical system behind the scenes and giving every plumber the same homepage with a different logo.
The technology should give us flexibility. It should not take it away.
A website should feel like the business it belongs to
This is probably the simplest way of explaining our position. When somebody arrives on a trade website, we want them to feel as though they have arrived at that company's website.
Not a plumber website.
Not an electrician website.
Not a tradesman template.
Their website.
The services should be theirs. The tone should suit them. The hierarchy should reflect what matters to their customers.
Where suitable customer photography is available, the website should make good use of it. The content should describe how the actual business works rather than how a generic trade business might work.
That doesn't mean every website needs to be wildly unconventional. Often the best design is straightforward. But straightforward and generic are not the same thing.
So are we anti AI?
No. This isn't an argument that AI is useless or that businesses should somehow avoid modern technology. That would be ridiculous.
AI is a tool. And, like most tools, what matters is how it is used.
Our objection is not to technology making web development better or more efficient. Our objection is to replacing the thinking, planning and expertise behind the website simply because software can generate something faster. There is a significant difference between using tools while building a website and handing the website over to a generator. For us, the important decisions still sit with people.
Could we build websites faster using AI?
Absolutely. We could probably build more of them too. But Toolbelt Websites wasn't created to produce as many websites as possible. It was created to build good websites for tradespeople.
- Websites that are structured around the individual business.
- Websites that work properly across devices.
- Websites that can be changed.
- Websites that can grow.
- Websites where the important decisions have been made deliberately rather than generated because they were statistically likely to look right.
Going back to that conversation with the plumber, AI can certainly get the nut started.
We would still rather use the spanner.
