Myth: a full redesign will lift conversions

Myth: a full redesign will lift conversions

Client wanted the big bang. New homepage, new everything, 'modernize it.' Six weeks of work.

We A/B'd old vs new. New design: -11%. Eleven percent down.

Problem with redesigns: you change 40 things at once. When it loses you have no idea which of the 40 killed it. No learning, just a crater.

We rolled back, then isolated. Tested the new nav alone (+3%). New hero alone (-9% — there's the culprit). Kept the nav, ditched the hero.

Redesigns are how you lose money with confidence.

Got a redesign queued? Break it into single-element tests. Ship the winners one at a time. Report back.