EcoPlate is a 96 well microplate with three technical replicates of 31 response wells with different sole carbon substrates + 3 control wells. The color development pattern of these wells obtained by incubation represents a metabolic fingerprinting, which is further processed for quantifying community-level multifunctionality and functional composition. However, the color fluctuations during the incubation and the color variations between replicates undermine the statistical assessment to functionally distinguish microbial communities. We developed a statistical protocol to improve the statistical power with two approaches. First, we tried to optimize the data treatment on color development during incubation and technical replicates. Second, we developed novel dissimilarity metrics by incorporating information of pairwise chemical dissimilarity among 31 carbon substrates into the computation. We implemented the protocol in a R script and tested it using data from microcosm and field experiments. We demonstrated substantial improvement of statistical power when using all measured temporal patterns during incubation compared to the case when using the pattern at the end of incubation only. We also found that the new metrics incorporating the chemical structural information improved the statistical power.