I think users do care about things like being able to use different apps with the same messaging service; non-techie friends of mine complain all the time about how some service X has a great mobile app but a bad web app and how service Y is the exact opposite. The problem with XMPP was it never really seemed to spawn the cool apps that would get the attention of non-techies (for whatever reason, but the fact you couldn't talk it natively from the browser can't have helped).
(Then again, I'm not exactly unbiased given I work for http://matrix.org)
XMPP, as a standard, was too slow to adopt mobile, so the cool apps had to invent their own messaging protocol (perhaps on top of XMPP) to get push notifications and such.
All they had to do is release their protocol specification, kind of like they did with SPDY.
It's quite obvious that when they started they were small compared to AIM, MSN, YIM and others so they used XMPP to help gain traction, once they became huge they dropped interoperability to lock users inside.
I can't understand this argument. There was not much extra stuff needed for mobile and most of the specs for it were written long ago. I suspect more people liked reinventing this easy-seeming wheel rather than XMPP was not ready.
I'm running Xabber on my phone and I refuse to use WhatsApp. Why would I pay for something when Xabber is free? WhatsApp is also a pain to install and verify your number if you don't have texting.
Oh yes it's early days for Matrix still. We do have clients for web, android and ios which we use day to day; they're nothing too fancy at the moment, but they're slowly getting better.
Matrix is very much still in beta; people have been quite excited about writing their own server implementations but have decided to wait until the protocol (and documentation!) stabilize. :)
(Then again, I'm not exactly unbiased given I work for http://matrix.org)