The redesign trap
Most landing page redesigns underperform the page they replaced. The reason: the existing page accumulated learnings (winning headlines, working CTAs, tested layouts) that get thrown out wholesale. A redesign starts from zero and has to relearn everything the old page already knew. Iterative improvement compounds; redesigns reset.
When to redesign anyway
Three legitimate reasons. (1) Strategic shift: new audience, new positioning, new product category — the old page is selling the wrong thing. (2) Performance ceiling: you've A/B tested for 6+ months and conversion has plateaued, suggesting the local maximum has been reached. (3) Brand refresh: the visual identity has changed and the page is genuinely off-brand.
The right way to redesign
Build the new page in parallel, A/B test it against the existing page (50/50 split for at least 4 weeks), and keep iterating both. Don't switch wholesale until the new version has beaten the old by a margin that exceeds your test variance. Score both versions as a sanity check before launch.