Nevertheless, initial trials suggested that the original SPA method produced real ambiguities for assessor and assessed alike about what the categories addressed and how serious the problems were. For example, a study conducted by Joe Besselman titled "A Collection of Software Capability Evaluation Findings: Many Lessons Learned" (presented at NSIA in March of 1992) showed that there was sometimes a difference of two maturity levels between evaluation and assessment ratings of the same organization when assessed by the SPA and SCE method. Some of these discrepancies resulted from the fact that internal SPA assessment teams were not familiar with the categories guiding the process, and some arose because early SPA assessments did not require the presence of a licensed and often external Lead Assessor who could guide the assessment team in the same way as the head of an external SCE evaluation. The Besselman study concluded, however, that the benefits of the SPA method were such that the method should not be changed to be more like an audit because it would then lose the organizational benefits of the assessment process . Instead, the report ultimately prepared for two significant SPA alterations: the development of an articulated description (or model) of the software capability development process so that internal assessors would have clear categories to work with, and the introduction of SEI-licensed Lead Assessors who could keep internal assessors within the same judgmental guidelines as those used in an SCE.
In regard to the first of these developments, the SEI began working on spelling out the structure implied by the original questionnaire so that the purpose and criterion for each element would be clear to all. Concurrently, Watts Humphrey elaborated the underlying principles of both the capability maturity concept and the assessment method in his 1989 work titled Managing the Software Process , which proved to be a landmark both in the theory of software process improvement and in the theory of software assessment.