SQR: Balancing Speed, Quality and Risk in Online Experiments

12:00 pm - 12:20 pm

Objective: Guidance
Audience Level: Beginner/Intermediate
Session Type: Presentation

Controlled experimentation is widely adopted to accelerate innovations. However, how fast we innovate is limited by how we run experiments. Most experiments go through a “ramp up” process where we gradually increase traffic to the new treatment. We have seen huge inefficiency in that process. In this paper, we build up a ramping framework that effectively balances Speed, Quality and Risk (SQR).


, Principal Staff Engineer & Statistician, LinkedIn