"User-centred design" sounds like something that requires a research department. It does not. At its simplest it means designing around what people are actually trying to do on your site, rather than around what you would like to tell them.
That distinction turns out to be most of the difference between a site that generates enquiries and a site that looks nice.
Start with the jobs, not the pages
Most website projects begin with a page list. A better starting point is a list of what people come to do.
For a typical service business that is something like: work out whether you do the thing they need, work out whether you are any good at it, work out roughly what it costs, and find a way to get in touch. Four jobs, in that order, and almost every visitor runs through them.
Once you have that list, the page structure mostly designs itself, and the arguments about whether the About page should come before Services resolve themselves.
Research you can actually afford
You do not need a lab. You need to stop guessing.
- Talk to five customers. Ask what nearly stopped them buying, and what they looked for before they got in touch. This takes a week and is more useful than any persona exercise.
- Ask whoever answers the phone. Sales and support know exactly which questions come up constantly. Those questions are your page structure.
- Read your search data. Search Console shows what people were looking for when they found you, in their words rather than yours.
- Watch someone use the current site. Give them a task, stay silent, and watch. Three sessions will produce more insight than three months of opinions.
Clarity outranks cleverness
The most common failure on a business website is not ugliness. It is that a visitor cannot tell what the company does within a few seconds.
Your homepage headline should survive the harshest test there is: would someone who has never heard of you understand what you sell? "Transforming ambition into outcomes" does not pass. "Webflow websites for B2B and SaaS companies" does.
The same applies throughout. Name your services what your customers call them, not what you call them internally.
One clear next step per page
Every page should have an obvious primary action, and it should be the one that makes sense at that stage.
A case study page is not the place for a hard sell; the natural next step is another case study or a line about how you work. A pricing page is exactly the place to ask for the enquiry.
When a page offers five equally weighted options, most people take none of them.
Hierarchy, and the discipline of leaving things out
Everything cannot be important. If a page has six things competing for attention, the visitor's eye has nowhere to land, and the usual result is a scroll to the bottom and a bounce.
The hard part of this is organisational rather than visual. Every department wants their thing on the homepage. Someone has to decide, and the decision should be based on what visitors are trying to do rather than internal politics.
Tell people what is happening
Interfaces should respond. A button that has been pressed should look pressed. A form that is submitting should say so. A file that is uploading needs a progress indicator.
The absence of feedback is why people click "Send" three times and you get three enquiries from the same person.
The same applies to errors. "Something went wrong" helps nobody. "That email address is missing an @" is a design decision that saves a completion.
Accessibility is user-centred design
If you are designing around the people using your site, that includes the ones using a keyboard, a screen reader, or a phone in bright sunlight.
The basics cover most of it: proper heading order, labels on form fields, alt text on meaningful images, visible focus states, adequate contrast, and never using colour alone to convey meaning.
It is also, incidentally, good for everyone. Captions get used in open-plan offices. Good contrast helps outdoors.
Test with people, early and cheaply
Five people will find most of the problems worth finding. You do not need a script, a lab, or a statistically valid sample.
Give someone a realistic task — "find out whether they work with businesses like yours and get a quote" — then say nothing and watch where they hesitate. Hesitation is the data.
Do this on the prototype, not after launch. Changes cost almost nothing before the build and are expensive after it.
Keep going after launch
A website is not finished when it goes live. Look at what people actually do: where they land, where they leave, what they search for, which pages produce enquiries.
Then change one thing at a time, so you know which change did what.
Where we start
Every project we take on starts with what the site needs to achieve and who it is for, before anything gets designed. It is also why our first question is usually about your customers rather than your competitors.
Running a usability test for nothing
You need one person, twenty minutes and a screen recording. That is it.
1. Recruit someone outside your business who roughly matches your customers. A friend of a colleague is fine.
2. Give them a realistic task, phrased as a goal: "work out whether this company could help with X, and get in touch."
3. Say nothing. The hardest part. Do not explain, do not rescue them. Silence is where the findings are.
4. Watch for hesitation, backtracking, scrolling past the thing they needed, and any moment they say "I assume…".
5. Ask two questions at the end: what was confusing, and what would you have expected?
Do this with five people and you will have a prioritised list of everything wrong with your site, for the price of five coffees.
Questions we get asked
How do I know if my homepage is clear enough?
Show it to someone for five seconds, then ask what the company does. If they cannot answer, the headline is not working.
Do I need personas?
Rarely for a small business. Two or three sentences about who you serve and what they need is usually more useful than a document nobody reads.
We disagree internally about the structure. How do we settle it?
Test it. Two people arguing from opinion will never resolve it; five customers will.
If you are planning a site and want a second opinion on the structure before anyone starts designing, we are happy to talk it through.