We used to do it some years ago. Some of our very first clients were getting A LOT of deep customisation work and we received decent compensation for the effort. Here's what we've learnt...

- Custom work takes consumes massive amount of time, mostly spent for negotiation pre and post delivery. Clients rarely feel happy about paying for that time. 9 out of 10 projects wouldn't have a clear technical assignment.

- As the platform developer we're in a crossfire. Clients would expect version updates see more to play nice with the mod (after all BoonEx did both), but wouldn't badge on requests to modify assignment to prevent potential version update disruptions. One's "vision" is untouchable. Problems are on us.

- IF we price in all the time spent for the work, further adjustments in case of version updates and all potential implications that may arise due to conflict of interest (like, say, us developing a default feature that is "similar" to the custom mod), then the job cost would look unrealistic.

I guess my point is that it CAN be done, but by someone else. Otherwise clients (or we) end up paying for issues arising from getting the job done by the main platform developer. Good 3rd party customisation group may be able to figure it out much better. I think AQBsoft guys have been rather good so far.

And in general, I believe that non-modular, broad customisation of an off-the-shelf platform is a bad idea anyway. Cost is comparable with development of a re-usable module with slightly less integrated "feel", while future-proofing is nearly impossible. More extensions, templates and options that can be carried over to newer versions is what I'd be looking at. You may also see that reflected in... you know what. :)