A client once knew, with the certainty businesses reserve for things they have never tested, that PayPal was load-bearing. Conversion depended on it. Customers demanded it. Removing it was the sort of suggestion that got a meeting cut short. So we turned it off for a fortnight and watched. Nothing happened. Not a dip worth the name, not a complaint that survived a second look. The thing everyone knew was essential turned out to be furniture.
Best practice without testing is received wisdom, and eCommerce sites accrete received wisdom the way old houses accrete wiring. The trust badges, the countdown timer, the pop-up, the carousel, the promo code box: each arrived years ago with a rationale, and the rationale was never re-examined because the element was never questioned. Optimisation teams then spend careers improving these fixtures. Almost nobody asks the prior question, which is whether the fixture deserves to exist.
The brief assumes the element
The sharpest version of this I’ve met came in a pitch, not a project. A major sports retailer wanted its promo code box optimised: better placement, better prompts, more redemptions. The behavioural data told a different story. Hardly anyone used it, and a good share of those who did had left the checkout to go hunting for codes, returning with a discount the business never needed to give. The honest recommendation was not a better code box. It was no code box. The brief had assumed the element; the data didn’t.
The method is almost embarrassingly simple. Walk the journey and ask of each element: what do we believe would happen if this wasn’t here? Rank the answers by fear. Then take the cheapest of the frightening ones and turn it off for a fortnight, measured, reversible, outside peak trading. Subtraction is the only test that costs nothing to build. The build is deletion.
There are honest exceptions. Payment methods carry compliance and audience nuances that deserve care, a fortnight can miss seasonal effects, and some elements exist for reasons conversion data won’t show, accessibility being the obvious one. The discipline isn’t recklessness. It’s refusing to let “we’ve always had it” stand in for evidence.
Optimisation asks how to improve what’s there. Subtraction asks whether it earned its place. The second question is cheaper and asked far less often.
If nothing changes, it was never necessary, and you’ve just bought a cleaner journey, a faster page and a shorter roadmap for free. In my experience, nothing changes surprisingly often. That’s not a comment on the people who built the page. It’s what happens when a decade of best practice goes untested, one reasonable addition at a time.