I finished the last piece with a question I didn't answer.

Understanding purpose, tradeoff, and consequence tells you whether you've thought about all three parts of a decision. It doesn't tell you whether the thinking is any good. It looks the same for a bad decision as it does for a good one.

If someone looking at a page with me...the client, a colleague, the person paying, and they say "I don't like it." What happens in the next thirty seconds decides whether we're doing design or exchanging preferences. If my answer is "I don't like it either," we're both working from taste. If my answer is "what isn't it doing?", we've started working from judgment.

Giving taste its due

Let me be fair to taste before I spend the rest of the article distancing myself from it, because I’m not treating it as an enemy.

Taste is a detection system. It's what tells you a page feels cramped before you can say why, or that the type is fighting itself, or that something about the hero is off but you can't name it. You can't skip it and you can't fake it. You build it by looking at a lot of work and noticing what's good about it, which takes years and doesn't have a shortcut.

Taste is what notices the problem. It doesn’t tell you who the page is for, what the fix is, or what the fix will cost. Taste rings the bell and then its time for the next job, and it's a different job.

The first test: could I be wrong?

When something bothers me, I ask: could I be wrong about this, and would I know?

If I say "this blue is too cold," I can't be wrong. It's a report about my own nervous system. Nobody can check it. Nothing changes if it's inaccurate, because there's no such thing as inaccurate when it's a fact about me, correctly stated.

If I say "this blue classes when the car and button is this color and the button is the only path to inventory, so the button has to win the contrast fight," that’s something we can check.

Someone can measure the contrast ratio and tell me I'm wrong. Someone can look at the analytics and tell me nobody is clicking the button. I've made a claim about the world, and claims about the world can be checked.

That's the difference. Taste is a claim about you. Judgment is a claim about the visitor, the business, or the page, something outside your head that can be argued with.

The second test: does it survive being reversed?

A real reason narrows the options. A fake reason can be stapled to anything.

Suppose a client asks why the hero isn’t a video, and I say "a image creates a stronger emotional connection." That isn't a reason. It's a sentence that would support the opposite decision just as well.

Reverse it: "a video creates a stronger emotional connection." Also sounds fine. A statement that survives being reversed isn't doing any work.

If instead I say "video means this page carries has to load several megabytes before anyone sees a price, and most of these visitors are on a phone, on a bad connection," It has a number in it and a consequence attached to it, and it could be wrong. If it's wrong I have to go back and change my mind, which is the entire point.

After you give a reason, reverse it in your head. If the reversed version sounds just as reasonable, you haven't said anything.

The third test: does it predict anything?

A real reason predicts.

If I'm right that the hero image is what decides whether visitors stay, then when we put the Mustang up there, the click-through into inventory should go up. That's a prediction I can be held to. If I'm wrong, I'll find out, and finding out is how the next reason gets better than this one.

A preference predicts nothing. "I like blue" doesn't tell you what will happen. It has no consequence you could check in three weeks, because it isn't about anything that happens in three weeks.

This is the problem with reasons that can't be wrong the ones where, whatever happens, you were right. If the change goes well it proves the change was good; if it goes badly it proves the visitors weren't ready.

Where this gets hard: when the taste is good

Now let me put all three tests on something real, because the taste versus judgment question gets much harder when the taste is good.

SuClasico's owner has been looking at competitors. One of them has a beautiful site. Cinematic photography, enormous type, a lot of white space, the kind of thing that ends up in galleries and wins awards. He wants SuClasico to look like that.

My taste agrees with him completely. It is a better looking site than what SuClasico has now, by a wide margin. If the question were "what looks better," the answer would be obvious and we'd be finished in ten seconds.

But that was never the question. SuClasico is not successful when it looks premium. It's successful when it helps the right visitors find cars, sell or trade their own, request a valuation, and trust the business enough to make contact. Those are the four jobs. Looking premium isn't one of them, though it probably helps with the fourth.

So I look at the award winning site the way I'd look at a competitor's showroom. What is it doing well that we should learn from? The photography, obviously. The restraint. The confidence to let one car occupy the whole screen instead of filling every inch with content. I'll take all of that in.

Then I look at what it costs. That design is built around a full-width image that owns the entire first screen. On a phone, the first thing a visitor sees is a photograph and a logo, and the inventory is a scroll away. On a slow connection and a good number of people looking at classic cars are not on fibre that photograph is the heaviest thing on the page.

The enormous type that looks good on desktop has to survive at three hundred and sixty pixels wide without turning into stacked lines.

It’s not an argument against the design. It's an argument about what the design is for, and whether it's the same thing SuClasico is for.

My taste has nothing to do with it. What do we take, what do we leave, and what will it cost?

Which parts serve SuClasico's and which parts just look good?

What evidence is there?

  • What people type into a search engine before they arrive.
  • What the owner hears on the phone.
  • What's filling the inbox.
  • What the analytics say people do after they land.
  • What a handful of real people do when you put the page in front of them and say nothing.
  • What accessibility and performance requirements rule out.

Inventing a persona and designing for it is how you end up serving a person who doesn't exist, extremely well.

Numbers tell you where to look. They don't tell you what to do. If most visitors leave the vehicle page without requesting a valuation, the analytics have found a problem. However they haven't told you whether the form is too long, the price is wrong, the photographs are bad, or the visitor simply wasn't ready.

Taste is not a character flaw

I don't want to leave thinking taste is something you grow out of. It isn't.

The habit isn't suppressing your reaction. It's labelling it. When something bothers me, I try to say to myself: that's a reaction, not a problem. Then I go and find out whether it's a problem. Often it is taste trained by years of watching what works. But the moment I say it out loud to a client, it has to become a claim about their business.

Why this matters more than it used to

You can create a page that looks like the award-winning competitor in the time it takes to describe it. If your whole value is that you have good taste, you're now competing with something that has unlimited taste and no fatigue.

The reasons are getting cheap too, and this is the part people underestimate. Ask for a reason why the hero should be a photograph and you'll get one, fluent and confident. Ask why it should be a video and you'll get one just as fluent, in the same voice, with the same structure.

What isn't cheap is being the person who says: the hero should be the car, not the showroom, because most of your visitors arrive from a search for a specific vehicle, and I'm willing to give up the emotional hit of the showroom photograph to answer that question faster. That sentence has a name on it. If it's wrong, you're wrong, and somebody can hold you to it in three weeks when the click-through hasn't moved.

Taste is free now. Reasons are cheap. Being accountable for saying it, predicting what it will do, and then finding out whether you were right, is something, people will pay for.

A reason is real when someone can hold you accountable for it.