A great piece (as usual) from Eric Seufert on what to look for in mobile analytics. There's so much to agree with in this that it's probably simpler to just go read it yourself, but this line certainly struck a chord:
Successfully iterating during the development of a mobile app – especially in the soft launch stage – requires a robust feedback loop and a sense of product development workflow that allows data to be transformed first into insight and then into product improvements.
That's 100% the case. Numbers are OK for reporting up in the organization, but in and of themselves they do precisely nothing when it comes to delivering success. You need to change those numbers - and that relies on how effective that feedback loop is, and how quickly and easily you can turn knowledge or insight into concrete actions. That's what we in Swrve aim to do, which is why we constantly stress the active aspects of our platform rather than the analytics (which are awesome, but we just don't talk about them!)
I'd also be fully on board with the comments around industry benchmarks. There seems to be a whole movement dedicated to analyzing these figures, and when they are not misleading (for all the reasons Eric lists) they are irrelevant. As the article correctly states, either your LTV is greater than your cost of acquisition or it isn't. And aside from that, shouldn't you be always looking to maximize all your KPIs, no matter where you sit against some theoretical benchmark.
As for A/B testing - I'll let the comment I placed on Eric's page speak for me on that one.