> Tell your management that usage of v6.0 is as good as gone, and you are wasting resources in doing any work to keep support for it

How do you say this without any knowledge whatsoever about the type of app they work on, or their target audience? v6.0 alone is almost 2% of users based on that chart. In fact if you include everything before v8.1, you seem to get something like 8% of users. What if their app, say, provides the poorest people access to public transportation? Would you really just drop 8% of people on the floor in any scenario?

Exactly this. Even if your demographic is the same as the chart, dropping 2% of users can be the difference between profit and destruction. Maybe it's margins that you'll save but keeping those users, possibly bad reviews, maybe market share over a competitor, there's plenty of reasons supporting older versions is not a definitive "waste."

Is it a pain to support old SDKs? Yep. But that doesn't make it inherently a waste of time to do so.

You have to weigh in the quality improvement of the app that you can provide by not limiting yourself to old apis, and also the resources that you can spend on improvement now that you aren't spending as much effort doing support for old versions.

What is the demographic of people who visit sites that have statscounter tacked on? It doesn't feel like it's representative at all.

What management? I decide this kind of thing myself in that project. The app is open-source and for a nonprofit. I don't see much reasoning to drop Android versions above 6.0 because there aren't that many API changes that would make a difference for my case. It's not like supporting 4.x or 2.x, the "you gotta carry the reimplementation of a substantial part of the UI framework with your app" kind of annoyance. The app in question doesn't even use appcompat. The apk is around 3 megabytes.

Well then it doesn't sound like supporting v.6.0 is much of an issue for you in your particular case. I'm happy for you.

Actually, 7 and 8 added some major support for Java 8, which is quite useful, as it adds better support for NIO and lambdas.

But most of that can be done with polyfills and transpilers, which the android gradle plugin does automatically nowadays.