September 16, 2026 · by David Gilbert · 4 min read · Web & E-Commerce
Not every redesign I've shipped has been an improvement, and I think honesty about the ones that weren't is more genuinely useful than only ever sharing the wins. One particular redesign, which I was genuinely proud of at launch, actually made the client's real numbers measurably worse for a while, not better, and the reasons why taught me something I still actively apply on every project since.
What I Actually Built
A cleaner, more modern, more visually striking design than the client's previous site, with genuinely improved aesthetics by almost any conventional design standard. I was proud of the actual visual work, the client was initially happy with how it looked at launch, and the early feedback on appearance alone was genuinely positive across the board.
What Actually Happened to the Numbers
Enquiries dropped, noticeably, within the first couple of weeks after launch. Not catastrophically, but measurably and consistently enough that it clearly wasn't simple random noise in the data. That's a genuinely uncomfortable thing to watch happen on a project you were personally proud of, and it forced an honest, fairly humbling investigation into exactly what had actually changed.
What Had Actually Changed, Once I Looked Properly
In pursuing a cleaner, more minimal aesthetic, I'd made the actual call-to-action considerably less visually prominent than it had been on the older, admittedly less polished design. The previous site's call-to-action button had been, frankly, a bit garish by current design standards — and it had also been completely impossible to miss. The new, more tasteful, restrained version was easy to overlook entirely if you weren't already specifically looking for it.
Why This Was Genuinely Hard to See Coming
Every individual design decision, viewed in isolation, was a genuine, reasonable improvement by normal conventions — cleaner typography, better, more generous spacing, a more sophisticated, restrained colour palette. The cumulative effect of several individually reasonable decisions was a site that looked considerably better and, in this specific case, converted measurably worse, and no single decision on its own would have obviously predicted that overall outcome.
What I Actually Did to Fix It
I made the call-to-action visually prominent again, deliberately, even though it meant compromising slightly on the pure, restrained aesthetic vision I'd originally pursued. Enquiries recovered within a couple of weeks of that one specific change, almost completely. The lesson landed hard and stuck permanently: aesthetics and actual function need to be evaluated together, honestly, not aesthetics first with function as an afterthought.
What I Changed About My Actual Process Afterwards
I now explicitly test prominence and clarity of every key action on a redesign, deliberately, before launch — not just "does this look genuinely good," but "would a new visitor, who's never seen this site before, instinctively know exactly what to do next within the first few seconds of arriving." That specific question, asked explicitly and deliberately rather than assumed, would have caught this exact problem well before launch, and it's caught several similar near-misses on other projects since.
Why I'm Sharing This Story Publicly
It's genuinely tempting to only ever share the wins, and I understand completely why most people in this line of work mostly do exactly that. I think the honest failures teach more, both to me and to anyone reading this, and admitting one publicly costs me very little compared to what it might save someone else from repeating on their own next redesign.
The Broader Lesson
Beautiful and effective aren't the same property, and they don't automatically travel together just because a design looks more polished and modern. The best, most successful redesigns achieve genuinely both. When you're forced to choose, in any specific tight spot, function has to win, even when it costs you a bit of pure visual polish you were personally proud of.