Cookies are shrinking and privacy regulation are increasing. How do you keep our A/B testing robust? Most testing tools (like Optimizely, Adobe Target, or VWO) abstract away the math. This talk explains how the tools make decisions, helping you understand concepts like variance, standard error, and confounding variables. “Collect everything just in case” is no longer operationally sound so Matt presents a framework for collecting only what is needed to prove a hypothesis, which is crucial for compliance (GDPR/CCPA) and efficiency. Learn how to design experiments that are mathematically sound and privacy-conscious by being intentional with your data. You don’t need to track everything to learn something valuable.