We, as an industry, used to do a lot of UX research. Today's designers ignore all of it because they don't like what it says. Spacial file managers? Global menu bars? 3D buttons? Ew. Let's just make shit up.
In reality, most design work these days is shifting to research and usability testing. Any designer who doesn't consider these things is just not a great designer and has much to learn.
Nearest I can figure, all the tuning is based on "engagement", which is a metric of how long a user spends on a thing. That's a useful metric to tune for if you only care about shoving ads at people, but is pretty much exactly the opposite of what you want if you're trying to help people get shit done.
Precisely. If modern touch user interfaces are the result of any carefully-gathered metrics (a premise I find suspicious, but let's run with it), the metrics themselves must be orthogonal to the user actually being able to complete tasks in as short a time as possible. Efficiency, speed, and accuracy were the sort of metrics used in legacy UI design.
But you are right, it's likely that modern metrics are things such as engagement, and whether the user is doing actions that are high-value for the company, such as sharing content with their social networks.
I think it's more that measuring engagement is cheap: just add some metrics to your app. Measuring usability is expensive: you need to give large groups of users your thing and be physically present to see how they try to interact with it.
I would be fairly surprised if startups regularly tested their UX the way e.g. Microsoft (used?) to test some of their products: give it to a room full of old people who have never touched your app and ask them to do something with it.
Well I can assure you that despite your experience, that is not generally considered good design methodology in the greater world of design. I don't doubt that it happens though.
This link is a great example of bad design, though. They don't even have to design for infinite viewport sides like you would on the web, there's a limited number of screen sizes this UI will appear on. In my work it's common practice to view your design at numerous different sizes and in the past issues like this were something we would specifically look out for when you have content of dynamic height that includes calls to action.
The solution is usually to programmatically ensure it is never perfectly cropped or, ideally, include hints there's more content or you can scroll.
There may be more R&D and UX testing. But there's a hell of a lot more design going on, much of it small and one-off. And the research doesn't seem to be filtering through.
Too, as noted, the UX metric is frequently "engagement", which is not the same as usability.
