A few days ago, Google made the new compose default on Gmail. It went from a separate page to a popup that’s accessible from anywhere in Gmail. And for the vast majority of the time, it’s better: it’s quicker to get to and makes it easy to reference other emails while writing a new one. Unfortunately, for attaching an image (not embedding it inline) or doing some heavy formatting, it’s a huge step backwards and makes me want the old compose back.
I’m sure the data backed up the decision. Only a tiny fraction of all messages needed this additional functionality so why worry about it? The problem with this approach is that even taking into account the infrequency, the cost of the workaround is large enough to cause a usability problem for the power users. It’s akin to the old version of the iOS App Store that would close itself every time you downloaded a new app. Sure that was great when you only wanted to download a single app but it made every other scenario significantly worse.
In the rush to be data driven, we shouldn’t forget the actual users and what they’re trying to do. A data driven approach should be used to improve our understanding, not replace it. Otherwise, we run the risk of “nice to have” features replacing the “must have” ones.