New research sheds light on whether common approaches for online advertising measurement are as reliable and accurate as the 'gold standard' of large-scale, randomized experiments.