Nobody can describe the look they want. Everyone can point.

---

QUICK ANSWER

People judge how a website looks faster than they can read a word of it, so asking a client to describe a style in words rarely works. The reliable way to brief a designer is to choose between pairs that differ in one thing only, where both options are good. How it works, and a version to try yourself, below.

---

Ask business owners what their new website should look like and most will reach for the same words: modern, clean, professional, simple. Then show them two homepages side by side, and they pick one almost instantly, and can usually say why.

That gap, between what people can say and what they can see, is where most website designs go wrong.

You judge a website before you can read it

In 2006, researchers at Carleton University in Ottawa showed people website homepages for a fraction of a second and asked them to rate how appealing each one was. Some pages were shown for half a second. Others for 50 milliseconds, one twentieth of a second, far too short to read anything.

The ratings barely changed between the two. The study, published in Behaviour & Information Technology, concluded that visual appeal "can be assessed within 50 ms". Its title put it plainly: web designers have about 50 milliseconds to make a good first impression.

So the judgement about a look is fast, visual and mostly wordless. Asking for it in words means asking someone to translate an instant reaction into vocabulary they do not have, and then asking a designer to translate that vocabulary back into a look. Two translations, and the meaning leaks out at both.

The words for a look are the same words for everyone

Modern, clean, warm, premium, friendly. These words are not wrong. They are just shared by everyone, so they separate no one.

A private bank and a children's dentist can both honestly ask for "clean and friendly". They mean completely different things: one means calm, spacious and restrained; the other means bright, rounded and cheerful. A designer handed only the words has to pick one meaning, and picks their own.

This is not the client's fault. Describing a visual style precisely is a professional skill that takes designers years. Expecting a restaurant owner or a clinic director to do it in a meeting is asking the wrong person the wrong question.

Mood boards measure the wrong thing

The usual fix is the mood board: collect websites and images you like, and the designer works out what they have in common.

It helps, but it has a flaw. When you compare a beautiful photograph of a hotel lobby with a plain screenshot of a software company, you are not choosing between two styles. You are choosing between good photography and no photography. You are choosing between a big budget and a small one. The thing you actually preferred is hidden under everything else that differed.

Choose between pairs that differ in one thing

The way around it comes from how preference is measured properly: show two options that are identical in everything except one thing. Rounded corners or sharp ones. Dense or spacious. Bold type or light type. Warm palette or cool. Everything else, including the content and the quality of the rendering, stays the same.

Now a choice means something. If you pick the spacious version, you preferred space, not a better photograph. Repeat that across several rounds, each testing whatever is still least clear, and a clear picture of the look emerges from choices that each took a second. Elevates settles the look this way in every discovery session: eight rounds, each pair differing on exactly one axis.

Both options should be good

There is a second rule, and it is the one most questionnaires break. Both options in a pair must be desirable.

Ask "do you want your brand to feel exciting or dull?" and everyone says exciting. The question measured nothing. But ask "daring or reliable?" and the answer genuinely splits people, because both are good things to be. A law firm that picks reliable and a festival that picks daring are both right, and the designer now knows something real.

When every option is a flattering one, the client's choices finally carry information.

A handful of such pairs covers most of what a designer needs to know. Daring or reliable. Warm or precise. Playful or serious. Rich in detail or calm and spare. Established or new. None of them has a wrong answer, which is exactly why each answer is worth having. A business that sits at reliable, precise, serious and spare needs a very different site from one at daring, warm, playful and rich, even if both owners described their taste as modern and clean.

Show it with the client's own name on it

The last piece is small and makes a large difference. The options are far easier to judge when they carry the client's own name, their own products and their own words, rather than a stranger's logo. People react to their own business differently. They can picture it on their customers' phones.

It also changes the moment when the design is first shown. If the look was built from the client's own choices, the first design is not a surprise to approve or reject. It is recognisable, because the client made the decisions that shaped it.

Try it yourself before you brief anyone

You can do a rough version of this alone, in an hour, before you ever speak to a designer.

• Pick three websites you like and one you dislike, from any industry. For each, write one sentence on what exactly you like or dislike: the space, the colour, the type, the photography.
• Look for the one thing. If you cannot name it, find a second site that is similar except for that one element, and see which you prefer.
• Write your choices as pairs: spacious not dense, warm not cool, rounded not sharp. Those pairs are a better brief than any list of adjectives.
• Note who takes your photographs. Elevates treats photography as the biggest single driver of how a site looks, and it is the thing briefs most often forget.

Hand that page to any designer, and the first design will be closer.

What Elevates does with this

Elevates built this method into its discovery session, the one described in the previous article in this series. The client sorts twelve character cards by picture before reading their names, sets six sliders where both ends are good things to be, then works through the eight rounds of choosing between two versions of their own site that differ in exactly one thing. The look is then computed from those choices, not picked from a catalogue, and shown back to the client as a consequence of what they chose.

For a business briefing a website, this changes where the effort goes. Either your own hour with the steps above gives a designer enough to get close first time, and that costs nothing. Or the look matters enough to the business that it deserves a structured session, and the first design you see is one you already recognise.