In the context of power analysis in experimentation, "power" refers to the probability that the experiment will detect an effect if there is one. In other words, it's the likelihood that if a change or difference truly exists, the experiment will correctly identify it.

Power is directly related to the Minimum Detectable Effect (MDE), the number of days or exposures, and the allocation. The MDE is the smallest change in the metric that the experiment can detect. The number of days or exposures refers to how long the experiment is active and the number of users enrolled in it. The allocation is the percentage of traffic that participates in the experiment.

By adjusting these three variables, we can increase or decrease the power of an experiment. For instance, a larger allocation or a longer-running experiment can increase the power, making it more likely to detect a true effect. Conversely, a smaller MDE or a shorter experiment can decrease the power, making it less likely to detect a true effect.