It's not just mobile apps, it's everything. I write demo software and libraries to control sensors that are used in everything from university lab benchtop res

It's not just mobile apps, it's everything. I write demo software and libraries to control sensors that are used in everything from university lab benchtop research to heavy automation. You can never trust that the user reads a word of the manual (well the heavy automation guys probably do, because there's real danger involved), so I do my best to make it so in the demo software, there's one button you need to press to start communicating with the sensor, and very clear options from there to change the sample rate, save data to a file, etc. In the APIs we provide for them to do their own programming, ideally they only have to declare a new instance of a single object, with the necessary configuration data passed in as explicit arguments to the constructors, and all the communication methods are immediately accessible from that class' API. There's not a lot of extra new classes to hold configuration data, etc., because then the user doesn't know which fields are necessary to set in those objects. Sometimes that means you get more parameters than you might like in a constructor, but it makes it a lot easier for the user because they know exactly what you require from them.