Randomization units are crucial in experiments as they determine the level at which users interact with an experiment, and how reliably teams can measure the impact of changes. The choice of randomization unit closely ties to the set of metrics being measured, and its selection should ideally precede metric consideration. Randomization units affect what experiences users see, have a profound impact on the metrics that can be analyzed, and determine the potential scope of measurements for those metrics. Choosing an appropriate randomization unit is essential to ensure statistical validity and accurate measurement, as it aligns with the independence assumption required by statistical analyses. If not matched correctly, this can lead to clustering, causing inaccurate results and potentially leading to false positives or false negatives. The correct choice of randomization unit allows for consistent ascription of experiences within that unit, enabling accurate analysis and interpretation of metrics at different levels of granularity. Consideration of randomization units is vital when designing experiments to ensure the best possible impact on end-users.