CUPED

Controlled Using Pre-Experiment Data

A statistical technique used to improve the efficiency of A/B testing by reducing the variance in experimental results. It leverages pre-experiment data, such as historical information or baseline metrics, to adjust the outcome of a test. By accounting for factors that might influence results before the experiment starts, CUPED can decrease the required sample size and duration of an experiment while increasing the precision of results.

In practical terms, CUPED allows businesses to conduct experiments faster and with greater accuracy by stabilizing the experiment data and reducing noise. This is particularly beneficial when resources are limited, or the business needs faster insights to make data-driven decisions.

Benefits of CUPED:

• Faster Experimentation: Reduces the sample size needed, speeding up the time to achieve statistically significant results.
• Improved Accuracy: Decreases variability in results by adjusting for pre-experiment conditions.
• Optimized Decision-Making: Provides clearer insights, allowing for more confident decisions based on the outcomes of experiments.

CUPED can shorten the time it takes to run web and feature experiments in the context of experimentation platforms, enabling businesses to roll out optimized experiences faster.

• Abbreviation: CUPED