We work on lot of AB test activities and use Analytics as the reporting source. There are times when we compare the expected vs actual test traffic in Analytics dashboard. Most of the time we see a variation of 10% in expected vs actual test traffic. Is it expected? What is the variance limit?
For example, if 5K users visited ABC page from July 1 to July 31, we expect that an experiment running on ABC page for all visitors during the same period should have 5K users as well.