Monetate asks the question, "How many observations are required to conclude that the observed difference between experiment and control are not due to random chance at least some defined percent of the time?" This calculation considers population variance that comes from the control group in an experience to compensate for the lack of preceding knowledge of population behavior that you may have.
Sampling error may occur when the sample contains actors that aren't part of your customer population (for example, an internal help desk that can place orders on behalf of customers, wholesalers that make repeated purchases, or bots that perform automated site testing). For these reasons, Monetate makes provisions and lets you remove identifiable actors that aren't part of the experiment population through the use of Stealth Mode.
Refer to Manage Stealth Groups for more information on excluding IP address ranges to prevent unwanted influence on your experience data.
Monetate calculates sample size requirements independently for each metric depending on what representative component of the population can be considered part of the sample.