The analysis of active learning is an important theme in technology assisted review research. The purpose of the study is to create a measurement to determine the efficiency of active learning algorithms, used in technology assisted review, in eDiscovery.The study is based on the research of Cormack & Grossman, (2016) and Quartararo, Poplawski, & Strayer, (2019), and conducted through a multiple case study by Yin, (2018). The findings show that current measurements of when an active learning algorithm can be determined as finished have troubles, if no control set is available. Hence, a new measurement of relative richness is proposed. Furthermore, two commonly used programmes, Relativity and Brainspace, are analysed with a newly created measurement of efficiency to compare their algorithms. Thus, Relativity has an advantage in efficiency compared to Brainspace. Organisations can use the measurements to test the efficiency between different algorithms and test their own active learning algorithms on when more training does not yield proportionate increase in precision, recall and F1.