If you're running experiments under pressure, the hardest part isn't ideas. It's bet sizing: deciding how big a bet to place based on expected value, and how much traffic to risk.
I size most of my bets with revenue per session (RPS), because it forces a clean link between an on-site change and dollars. For bet sizing, conversion rate alone can lie to you.
This is my practical way to do experiment bet sizing when time, traffic, and patience are all limited.