The problem is fundamentally perception. Once they bring the extreme outliers back into line, then it’s a matter of encouraging diversity through other means. The problem, as I believe you’re aware, isn’t so much that some troops perform better than others (variance in defense win percentage is very small, regardless of defense team composition) but that most people don’t want to think about their team composition (which is a real shame, considering that’s the real meat of the game), and just go with whatever they’ve heard is the best.
The game just needs to encourage people to try more troops. I’m a bit surprised that the event system, with its rotating buffs, hasn’t done a better job of that. But there’s also the resource investment to consider – how many people can just level and trait a new team (or several, as they experiment) every week? Again, people would rather not have to think about it, and just look for the dominant strategy – not because it wins more reliably, but because of the effort/reward ratio. If unique, creative (defense) teams were more capable of catching players by surprise and actually affecting the win/loss ratio, people might feel more encouraged to experiment. Unfortunately, the game caters to speed and time invested over strategy, which is largely why I’ve stopped playing.