As an opposing datapoint, I actually had an unexpected A/B result when testing the recent S3stat redesign. Trial signups for the new version were better, but o

As an opposing datapoint, I actually had an unexpected A/B result when testing the recent S3stat redesign. Trial signups for the new version were better, but only by a percent or two. Not really enough to bother switching. But I also tracked all the way through to purchases (which typically happen after 30 days of use), and I'm glad I did. After two months, the people seeing the new version were converting twice as well as the ones seeing the old version. As in, double the sales. As in, wow. So I guess it's entirely possible that users do in fact respond to pretty design. The only thing one can do is test and see.