I find I'm in the same end-user boat. I've used KDE since I don't know when and I used to recommend it to anyone who would listen but it's got to the point that even though I mostly still use KDE I would never offer to install it for anyone else again. I am almost desperately looking for a suitable Qt replacement, have been for a couple of years. The only things that keep me using it are Kwin and Akregator. The main argument I used to defend my obsessive use of KDE was Kmail but I swapped that out for Thunderbird a year ago and with it's conversation add-on I don't think I'll be going back to Kmail. Also, I've got an 8Gb i7 laptop and it's not enough for long running KDE and apps, it's need at least 16Gb ram to be able to leave all apps running all the time without swapping. That's crazy.
My older sister is a classic case, she depends on me to help her with her laptop (hence windows is not an option) which has had KDE since about v4.4 so I've had a good opportunity to witness a novice user deal with it and I have to say, the promise of a usable and stable OS with a future, has not eventuated. Every update creates a different set of small problems that a non-techie has no hope of solving. That's partly distro related (Kubuntu) and partly KDE at fault but from her point of view it's that "linux thing" is not working again whereas all her friends Windows7 laptops "just work", and never crash these days.
From an outsiders non-dev point of view: if KDE user and developer adoption is not growing at a significantly faster pace than it has been for the past 2 years then the core team should have a good long look at their strategy and consider cutting the OS in half, isolate the core that is really important, eject the rest to playground, and seriously focus on getting the core working 100% and not waste any effort on non-essential part. The OS is simply too big for the number of people involved. I feel the premature run to v4.0 so impacted end-user adoption that the overall project has never recovered, and maybe never will. So sad for my favourite desktop.
I personally want a HTML5 desktop, something I feel I can modify and have control over. Perhaps a Qt5/QML hybrid based on webOS in September.