The crux of the SEI-based assessment reference models is a vision of how organizations stabilize themselves so that random efforts toward improvement can evolve into structured, reliable, and continuous building of strength upon strength. For both the CMM and the staged version of the CMMI , five capability maturity levels are posited: Level 1 represents a condition in which processes are unarticulated and improvements are random and sometimes contradictory. At Level 2, project management processes are stabilized and articulated so that technical developments can be approached in a predictable way. At Level 3, the best project and technical processes are identified and institutionalized in an organization-wide platform so that the organization can centrally support improvement efforts, including training. At Level 4, both projects and the central organization begin to use baseline measurements to compare the strengths and weaknesses of past and current processes and products. At Level 5, the organization and the projects are able reliably to anticipate risk and bring in new technology with a firm grasp of the consequences of change, and to initiate programs of continuous improvement in a systematic and measured way.
In short, assessments analyze not just whether organizations perform functions well but also, in reference to a process improvement model, whether they are likely to reliably generalize lessons of continued and increasing excellence out of local successes.