The redesign that made things worse.

A pattern we see often enough to predict: a company invests heavily in a redesign, launches to internal enthusiasm, and watches conversion stay flat or drop.

Nobody can explain it. The old site was dated. The new one is clearly better. Everyone who saw it said so.

The explanation is usually in the brief. It was never about conversion.

What redesign briefs actually ask for

Look at how these projects get scoped and the language is almost always about perception rather than behaviour. Modernise the brand. Bring it in line with competitors. Make it feel more premium. Clean it up.

Every one of those is a statement about how the site should look, evaluated by people inside the company.

Almost nobody briefs: here is the sequence of questions a buyer needs answered before they will hand over a card, and here is where the current site fails to answer them.

So designers optimise for what they will be judged on — which is how it photographs in a portfolio and whether the client likes it in the presentation. Those are legitimate professional incentives. They are just not the same objective as selling.

Where aesthetics quietly cost money

Good design and good conversion are not opposed. But they are not the same thing, and when they are not explicitly aligned the aesthetic wins by default.

Some concrete, extremely common examples:

Generous whitespace that pushes the offer below the fold. Whitespace is a real design virtue. It is also how a hero section ends up requiring three scrolls on a phone before anyone sees what you sell.

Subtle, low-contrast buttons. A call to action that blends elegantly into the page is a call to action people do not notice. Contrast is the entire job.

Lifestyle photography replacing product information. Nocs Provisions sold binoculars and led with atmospheric outdoor imagery. Beautiful, on-brand, and wrong — optics buyers want specifications and comparisons. The most likely purchasers bounced first.

Navigation designed for completeness. Every page in the header is a menu nobody can parse. The goal is to get people to a decision, not to demonstrate that the site has depth.

None of these are mistakes in a visual sense. Each is defensible in a design review. They cost money anyway.

The four questions, in order

The framework we use before any layout exists. A visitor has to get satisfying answers to four things, roughly in this sequence:

1. What is this? Within a couple of seconds. Not your positioning statement — the literal category. Clever headlines routinely fail here, because cleverness requires context the visitor does not have yet.

2. Is it for someone like me? People need to recognise themselves. This is what social proof, use cases, and specificity do — not general credibility, but she has my problem.

3. Why should I trust you? Reviews, guarantees, real photography, a physical address, named people. Anything that signals a company that will still exist next month.

4. What happens if I am wrong? Returns, warranty, sizing help, how to reach a human. Miss this and you lose everyone risk-averse, which in most categories is the majority.

Miss the third and you lose everyone who has been burned before. Skip straight to the offer and you lose people who never worked out what they were looking at.

Most underperforming pages answer these out of order, or answer one at length and skip two.

Design mobile-first, and mean it

Most e-commerce traffic is mobile. Most conversion loss is mobile. And most sites are still designed on a wide desktop canvas and then compressed.

The result is predictable: a hero that works beautifully at 1440px and requires four scrolls at 375px to reach a buy button. Copy sized for a column that no longer exists. Tap targets built for a cursor.

Designing the constrained view first forces the hard prioritisation decisions early — because on a phone you genuinely cannot show everything, so you have to decide what matters. Then let the desktop layout expand from that. It is a better brief in both directions.

Build for iteration

The last one, and it is structural rather than visual.

Sites that require a developer for every copy change stop improving the day they launch. Someone notices the headline is not working, files a ticket, and it enters a queue. Four months later it is still queued.

Which means the version that converts best is a version you never got to.

Build so the team can edit — copy, images, section order, offers — without filing a ticket. The compounding gain from a hundred small improvements over a year exceeds whatever you gained from the redesign itself.

The alternative to redesigning

Most sites do not need a rebuild. They need work on the handful of pages that carry traffic and money.

A full redesign is slower, riskier, and resets the data you were learning from. It also tends to change everything at once, which means when conversion moves you cannot tell what caused it.

Start with the pages people land on and the pages money passes through. Fix the decision path on those. If the platform is genuinely holding you back or the structure is beyond repair, rebuild — but that is a smaller share of cases than the redesign industry would suggest.

And whatever you do, plan the migration before design starts. URL changes without redirects, removed content, and lost internal linking are the most common way a redesign causes lasting damage — and the one nobody discovers until traffic drops the week after launch.