Whereas there exists a considerable amount of literature on the final challenges and finest practices for reliable on-line A/B testing, there are restricted research on pattern measurement estimation, which performs a vital position in reliable and environment friendly A/B testing that ensures the ensuing inference has a ample energy and kind I error management. For instance, when the pattern measurement is under-estimated the statistical inference, even with the right evaluation strategies, will be unable to detect the true vital enchancment resulting in misinformed and expensive selections. This paper addresses this basic hole by growing new pattern measurement calculation strategies for correlated knowledge, in addition to absolute versus relative remedy results, each ubiquitous in on-line experiments. Moreover, we handle a sensible query of the minimal noticed distinction that will probably be statistically vital and the way it pertains to common remedy impact and pattern measurement calculation. All proposed strategies are accompanied by mathematical proofs, illustrative examples, and simulations. We finish by sharing some finest practices on varied sensible matters on pattern measurement calculation and experimental design.