Smart approach honestly - cap the downside first, figure out the model later. That's basically what we did at DataSift early on too.

One thing worth thinking about as you iterate: track the delta between what customers would pay for the outcome vs what it actually costs you in compute. That gap is your real margin, and it'll shift fast as model costs drop. The companies that instrument that well early will be able to move to outcome pricing confidently when the time is right.

Enjoy the holiday - sometimes the best product thinking happens away from the keyboard.