Redesigns Fail More Often Than Tests Do

A full redesign changes hundreds of variables at once, so when revenue moves you cannot tell which change did it. If it drops, you cannot roll back one piece. Iterative testing is slower per change and far faster at producing knowledge, because every result tells you something you keep.

The problem is attribution, not aesthetics

New sites often look better. That is not the issue. The issue is that a redesign is one enormous untested change, and the base rate for untested changes is poor.

Optimizely’s analysis of more than 127,000 experiments found roughly 12% produce a statistically significant positive result. Most changes teams believe are improvements do nothing or actively hurt. A redesign bundles hundreds of those bets together and ships them on the same day with no way to separate the winners from the losers.

What you lose when it goes wrong

Six months of work and a revenue decline you cannot diagnose. The usual response is a partial rollback based on opinion, which restores some of the old problems and keeps some of the new ones.

What you also lose is the learning. A test that loses tells you something true about your customers that stays useful for years. A redesign that loses tells you only that the redesign lost.

Where redesigns are the right call

They exist, and pretending otherwise is dishonest.

Platform migrations, where the old system genuinely cannot support what the business needs. Brand repositioning, where the visual system has to change for reasons that have nothing to do with conversion. Technical debt severe enough that every test is blocked by the build. And sites so far behind on performance that incremental fixes cannot close the gap, which is not rare given only around 42% of mobile sites pass all three Core Web Vitals.

How to redesign without flying blind

The traffic split is the one most teams skip and the one that matters most. Rolling a new site to half your traffic for three weeks costs a little complexity and buys you the ability to reverse a mistake before it has cost a quarter.

The constraint that improves the work

Working inside an existing design forces better hypotheses. When you cannot rebuild the page you have to find the specific thing that is failing.

At a personal care brand we worked with, brand guidelines ruled out almost every layout change we would normally reach for. The programme ran on merchandising, navigation and post-purchase instead, and post-purchase upsells alone added around 10% to average order value. That result came from a constraint, not despite one.

What twelve months of iteration produces

Not a dramatic before-and-after screenshot. A sequence of small confirmed gains that compound, plus a body of knowledge about what your customers respond to.

Research from Harvard Business School cited in industry analysis found companies adopting systematic A/B testing see performance improvements of 30% to 100% within a year, with the highest testing velocity compounding fastest. That is a year of many small decisions, not one big one.

The question to ask before signing off a rebuild

What specifically is wrong with the current site, and how do we know?

If the answer is a list of evidenced problems, ask whether each one could be fixed and tested individually. Usually most of them can, for a fraction of the cost and with a result you can actually read.

If the answer is that it feels dated, that is a brand decision. Make it as a brand decision, budget it as one, and do not expect it to pay for itself in conversion.

Then ask one more thing. What is the plan if the new site converts worse than the old one, and who has the authority to call it? A rebuild with no rollback decision agreed in advance is a rebuild you will defend rather than measure.

More on how we structure conversion rate optimization services for DTC brands, including how we handle sites mid-rebuild.