Personally, I wouldn't shy away from A/B testing prices (I do it), if you have enough transactions going on to run a successful experiment.

Running pricing cohorts takes more time, and doing other experiments at the same time will taint the pricing experiment.