CUPED Methodology

CUPED stands for Controlled-experiment Using Pre-Experiment Data. It is a technique that reduces noise in an experiment so you can learn faster. It uses information about each visitor from before the test started, such as past purchases or visits, to adjust the analysis. When noise is lower, you need fewer visitors to see a real effect. The method keeps the result fair and unbiased while giving you a clearer picture. Example: Use each person’s pre‑test browsing time to clean up your signup experiment so a small improvement becomes visible sooner.