There are also issues of training and qualification across types. If you make a cockpit too different to what's gone before, you'll make it harder to learn for those with experience - and increase the likelihood of mistakes even after type qualification. And, if you're selling into a large carrier with a lot of aircraft you've sold them before, you'll lose a big advantage if you make your new type too different to the old. The other chap can make the argument that if the carrier is going to need a lot of retraining anyway, why not get something really good? And new is always good, right?
I learned this in the distant past when I used to go to Pprune meets, because it's always bothered me that so many aspects of flight decks (especially the mundane bits that only really matter when things go wonky) are clearly awkward hangovers from the past and I've always been a passionate fan of usability. It was one of the first questions I asked over a beer, and that answer was one of the first real lessons for me that usability is far more complex than just being logical about things. Humans, even pilots, depend on many factors to do their jobs well, and you've got to really understand humans before you have a chance with human factors.
R
(The fact that pilots were human too was also something impressed on me deeply from those Pprune nights. But that's a different story...)