Seems like there's a lot of parallels here between a scrum style sprint and what you're doing, except the iteration is muuuuch longer.
(ie: setting the iteration for a fixed length of time, ideally freezing the features in the sprint, post sprint review, etc).
One wonders if you could either take more practices from scrum to make this better (90 days seems like a really long sprint cycle), or if lessons from your 90DP execution could make your development practices better.
We had to go with 90 days because it gives us enough time, company-wide, to execute. Anything shorter and you're constantly changing course and distracting everyone.
Interesting. Might be something to evaluate at the end of the 90DP - sounds like you are assigning too big tasks and doing waterfall style work. Maybe it works for these things, but seems to me, "Can we iterate fast enough with a 90 day cycle? How could be cut that in half for the next iteration?" would be my first retrospective question.
(ie: setting the iteration for a fixed length of time, ideally freezing the features in the sprint, post sprint review, etc).
One wonders if you could either take more practices from scrum to make this better (90 days seems like a really long sprint cycle), or if lessons from your 90DP execution could make your development practices better.