I think there are a few related causes behind this behavior: 1) Visual changes show better than feature lists. I'm not convinced this matters anywhere near as

I think there are a few related causes behind this behavior: 1) Visual changes show better than feature lists. I'm not convinced this matters anywhere near as much for customers as marketing folks seem to think it does, but marketing values it regardless, management values it (it'll show better to their bosses, too), and product managers, designers, and front-end developers value it (so they have nice-looking things for portfolios and presentations, both internal and external) 2) Making something that looks good in a screenshot or demo is not the same thing as making something that works well (that is, has "good UX"), and making something that does both well is hard enough that the talent (and environment/budget) to do it may be fairly rare. The former matters to those concerned about point 1, so that's what gets priority (which is why so many redesigns are neutral, at best, as far as UX, and many are worse). 3) If design is done well, it's heavily cyclical. Design guidelines and languages and component libraries and such should be built up such that most design work only happens when there are initial designs, very large feature add-ons that break with what the previous design guidelines can accommodate, or redesigns. So, if you're a designer and want to stay comfortably employed, and you're a manager who doesn't want to surrender head-count or lose your people to other projects, and a large part of the org has various incentives to push initiatives that result in something nice to to look at (see 1, again) what happens? Lots more redesigns than need to happen. LOTS more. [EDIT] And all this is before we even consider possible business reasons to deliberately make the product worse, which are also A Thing.