Make it pop

A client asked me for a homepage carousel once and I did the thing I’m not proud of, which is I said sure, and then I built a carousel. Nice one, too. Swipe gestures, reduced-motion respected, the whole bit.

Nobody clicked past the first slide. Depending on whose numbers you trust it’s something like 90% of the interaction landing on slide one and the rest being statistical noise, and ours was worse than that. Which I could have told them going in, and sort of did, in the mealy-mouthed way you do when you’ve already decided to build it anyway.

The part I missed was why the carousel existed at all. It turned out five departments each wanted the homepage, and there was no version of that meeting where four of them lost gracefully. The carousel was the only shape that let everybody keep their slide.

People hand you the answer, not the question

I think this is the actual job and it took me an embarrassingly long time to notice. Requests almost never arrive as problems. They arrive pre-solved. “Can we add a filter here,” “can this be a dropdown,” “can you make the logo bigger.” Somebody has already done the translation from what’s bothering them into what they think you do, and the translation is where all the information got lost.

And they’re not being cagey about it. They’re being helpful. It feels rude to walk into a room full of professionals and describe a vague feeling, so people tidy it up into an instruction first. Half the time they’re apologizing while they do it.

“Make it pop” is the canonical one and it gets mocked constantly, which I find a little unfair. It’s not a taste problem, it’s a vocabulary problem. They can see that something’s wrong. What they don’t have is contrast, hierarchy, whitespace, type scale, the words. Expecting a client to show up with those words is basically expecting them to do the work and then pay you to type it in.

The carousel wasn't a design decision. It was a peace treaty, and I'd been hired to ratify it.

The part where this curdles

Here’s where I want to be careful, because the “clients don’t know what they want” genre is about one inch away from “so ignore them,” and that’s a much worse failure mode than building the carousel.

The client knows things you structurally cannot know. They know the customer who calls every Friday. They know which invoice field exists because of a lawsuit nobody talks about. They know that the thing you want to delete is used by exactly one person who happens to control the renewal. When they push back on your genuinely elegant idea, the base rate that they’re simply wrong is way lower than it feels in the moment.

So the move isn’t treating the request as noise. It’s treating it as evidence. Somebody doesn’t ask for a carousel out of nowhere. The carousel is a fingerprint that a real problem left behind, and your job is the problem, not the fingerprint and not your opinion of their taste.

Questions that actually get you somewhere

For what it’s worth, the ones that keep earning their keep for me:

“What happens if we don’t do this?” The answer is either “nothing, I guess” (congratulations, it wasn’t a requirement) or a very specific name and a very specific bad afternoon, which is the whole brief.
“Who asked for this?” Requests get laundered. By the time it reaches you it’s been through a PM, a stakeholder and a forwarded email, and the original sentence was different.
“Tell me about the last time this bit you.” Past tense over future tense, always. People are unreliable narrators of hypotheticals and extremely reliable narrators of grievances.
“Can I just watch you do it?” 20 minutes of screen share beats an hour of interview. What people describe and what people do are two different datasets.

And then the slightly underhanded one: show them something wrong on purpose. Not sabotage, just a real option you don’t love, sitting next to the one you do. People are mediocre at describing what they want and outstanding at rejecting what they don’t, so give them something to reject. Two rounds of that gets you further than any amount of asking.

The late request is usually the real one

You know the one. Week nine, everything’s built, and the email starts “oh, one small thing.” And the instinct is to treat it as scope creep, sometimes correctly.

But a lot of the time that thing didn’t come up earlier because it was too obvious to say out loud. Nobody tells you water is wet. Nobody mentions that every order has to be approved by the branch manager before it ships, because of course it does, that’s just how orders work, that’s not a fact, that’s weather. It only surfaces when the thing you built collides with it.

I’ve started reading the late request as a gift rather than an ambush, or trying to anyway, with mixed results 🙃. It’s the first time somebody has described their actual world to you instead of their idea of your world.

Being right is worth nothing

The worst outcome in this job isn’t building the wrong thing. It’s building exactly the thing that was asked for, watching them be quietly disappointed, and being completely in the right about it. You can go get the email. The email will say what you remember it saying. Being right is worth approximately nothing at that point, and you still have to fix it.

The reframe that helped me most is that a client being “difficult” is usually a client describing something they can see and you can’t. That’s not obstruction, that’s just the information being unevenly distributed, which is the normal state of every project ever. The hard part isn’t even asking better questions. It’s being someone they’ll answer honestly, because the real reason is often political or embarrassing or “my boss said so,” and nobody leads with that on a kickoff call with a stranger.

All of which assumes you can get the actual decision-maker on a call, and I’m aware that’s a small-and-mid-market luxury. Six layers deep in an enterprise procurement process, I don’t know, I’m sorry, godspeed. Ask what happens if you don’t do it anyway. Somebody in that chain knows, and they’ll be relieved somebody asked.