A Twitter thread by Everard Hunder on gratuitous tinkering.
« 1) If a new website generates less sales (lower conv rate) than the old website, you’ve failed. You have a huge job ahead to understand what you’ve changed, why you changed it, and to reconcile the old elements that converted versus the new elements that don’t.
2) A methodology. Before any website project you need to understand which existing headlines, copy, pages, images, devices, user flows are converting and working hard. These are the untouchable bedrocks. Change these at your peril.
3) Use the old website as a testing ground for your proposed new elements – design, copy, headlines, images, user flows. Confirm these new elements outperform the old. Most won’t. A/b tests, multivariate tests, new landing page tests, new forms etc.
4) Use data to drive website improvements, not personal opinion and ‘design’ thinking. The outcome is more sales, more conversions. This methodology means your chance of beating the old is website is significantly higher than a one shot guess big bang new launch. Good luck.
Sometimes those artistic and aesthetically perfect sites with immaculate white space DO outperform a daggy old site. But that is a high stakes game risking all on some overpriced html wallpaper. »