This paper reviews the purposes of CIPM key comparisons and presents a model-based analysis that is in keeping with these purposes. Measures of ‘operability’ and ‘interoperability’ are derived to answer the reasonable questions ‘how accurate will the result be when the laboratory in question makes some measurement in the future?’ and ‘how similar would the results be if two laboratories were to make a common measurement in the future?’ These are explicit questions that relate to the concepts of measurement ability and the equivalence of standards. The basic concepts emphasized are that (i) a measurement result cannot legitimately contribute to its own point of reference and that (ii) the effects causing error, or uncertainty, that are relevant in forming the measures of operability and interoperability are those that vary after the key comparison. Therefore, points of references for each measurement result are formed by excluding the result in question, and the measures of operability and interoperability are defined using the uncertainty components with effects varying on the short time-scale. These measures have clear and communicable meanings.