You're evaluating a feature in a low-traffic product area where user volume is limited and experiments take a long time to mature. You need to decide how to run tests when standard A/B setups may be underpowered.
How do you think about experimentation in a low-traffic product area?