I reached out to Attodé first. Not because the site was broken, but because of the gap between what it was and what it could be — a genuinely good brand running on a theme that looked like every other theme.
I went in with my eyes open. Fashion is one of the hardest categories in ecommerce to move with CRO. Nobody is comparison-shopping a jacket the way they compare a spare part: the decision is emotional, it’s driven by the product and the brand rather than by the page, and a big share of the traffic is browsing rather than buying. That flattens most of the levers CRO usually pulls.
Nobody buys a jacket because you moved the trust badges
People don’t buy a jacket because you rearranged the trust badges. They buy it because they want the jacket. The conversion lever is smaller than in almost any other category, and anyone who tells you otherwise is selling you a retainer.
So I never planned to run the usual programme here. Button colours, badge placement, a new microcopy line above the add-to-cart — the small stuff a lot of agencies bill for month after month — was never going to shift anything on this store. In a category this resistant, a half-percent nudge disappears into the noise before you can even read it.
It was big swings or nothing. Rebuild the parts of the store that every visitor actually moves through, make them look like Attodé instead of like the theme, and accept that some of it would be measured and some of it wouldn’t.
The constraint that shaped the whole collaboration
Two things were true at once.
First, the category resists testing. Second — and this is the one that actually shaped the work — a third-party app on their store breaks the design in draft themes. Theme-level A/B tests weren’t viable, because you can’t fairly test a variant that renders broken.
That ruled out the standard playbook. So I split the work in two: test what could be tested cleanly, and build what couldn’t — then decide on judgment, with the client, out in the open.
Here’s some of what came out of that. Not every test we ran for Attodé is public, but these are the ones we can show.
The product page moved. That’s the win.
Tested. On the product page the gallery became a continuous scroll instead of dots, I built a breadcrumb they didn’t have before and placed it at the foot of the page, and the size guide got a small clean-up. Over 32 days it returned +4.65% revenue per user and +3.6% conversion rate.
It’s not a statistically bulletproof result, and I’m not going to pretend it is. But in a category where most product-page tests come back flat, a page that reads more clearly and earns more per visitor is a page worth keeping. We rolled it out.
Tested. We also tried a horizontally scrolling submenu on mobile, to put more of the catalogue within reach. It lost: −8.5% revenue per user and −2.2% conversion rate over the same 32 days.
We kept it anyway, and that was a joint decision with every number on the table. The new submenu fits the brand far better than what it replaced, and we agreed that the design was a net improvement to how the store feels even though this particular test didn’t pay for it. Losing a test isn’t the same as losing an argument — it just means you make the call knowingly instead of blindly.
A completely custom menu and cart, shipped on judgment
The rest of the work couldn’t be tested at all, so we built it the way you’d build anything else you can’t A/B: carefully, against clear references, with the client in the room.
The navigation menu was the theme’s out-of-the-box default. We replaced it with a custom one on both mobile and desktop, built around what Attodé wanted their catalogue to feel like — far more customisable, and it finally looks like the brand rather than like the theme.
The cart drawer was close to empty. The new one carries a free-shipping progress bar, dynamic upsell items and trust icons, so the moment a visitor is closest to buying is doing some work instead of none.
Both went live without a controlled experiment, because the draft-theme constraint made a clean one impossible. That means I can’t tell you what either did to revenue, and I won’t invent a number for them. What I can tell you is that both are still live on the site today.
What we set out to do, and what we got
I knew going in that fashion would fight back, and it did. I’m still happy with where this landed.
We rebuilt the three areas that nearly every visitor actually passes through — the top of the product page, the navigation and the cart — on mobile and on desktop. Some of it is measured: the product page earns more per visitor than it did. Some of it isn’t, and I’ve said exactly which parts.
And the part that doesn’t show up in any test tool: the store stopped looking like a theme and started looking like Attodé. That was the point of reaching out in the first place, and it’s the thing they kept.
