The healthy ingredients of a process performance model are summarized below:
1.The model is statistical, probabilistic, or simulation based. This particular ingredient emphasizes the logical consistency of two CMMI process areas: Quantitative Project Management (QPM) and Organizational Process Performance (OPP). QPM stresses the need for understanding statistical variation of process performance factors. Additionally, QPM reinforces the need to separate assignable, special cause variation from inherent common cause variation to help understand what actions to take with respect to each type of variation. This healthy in-gredient emphasizes the need for process performance models to model the uncertainty of the predictive factors and their resulting impact on the uncertainty of the behavior of the outcome factor. For this reason, deterministic models that merely perform mathematical calculations on point estimates fall short of the superior information achievable from models that are sta-tistical, probabilistic, or simulation in nature.
2.The model predicts interim and/or final project outcomes. This ingredient derives more from practical experience and management’s need for real-time cycles of learning within a given project or program. To maximize real-time cycles of learning within a given project or program, managers need to predict interim performance outcomes in addition to the traditional end-of-project performance outcomes.
3.The model uses controllable predictive factors that are directly tied to subprocesses or work activities. This healthy ingredient focuses on the need for process performance models to be actionable. From that standpoint, if a model does not have at least one controllable predictive factor, it does not directly promote insight for action to influence the undesirable predicted outcome. For example, traditional project forecasting models that model only uncontrollable
factors make predictions that offer little help or insight into the actions to be taken to drive a more desirable predicted outcome. Additionally, this ingredient highlights the need for the controllable factors to be detailed enough to show a clear link to a specific subprocess or work activity. This clear link enables proactive management responses.
4.The model quantitatively characterizes and models the variation of the predictive factors and describes the predicted range, uncertainty, or variation of the outcome performance measures. This ingredient is a chief overlap of CMMI high maturity and Six Sigma concepts. Recogniz-ing that variation (i.e., risk) may very well be unbalanced and significant in the real world, the models account for this by modeling the uncertainty of the predictive factors. Numerous ex-amples exist in industry in which analysis using only the mean or average estimate rather than the distributional information caused serious problems in predictions of schedule, performance, and other modeled factors.
5.The model enables “what-if” analysis for project planning, dynamic re-planning, and problem resolution during project execution. This ingredient builds on language in the CMMI Organizational Process Performance (OPP), Quantitative Project Management (QPM), Organizational Innovation and Deployment (OID), and Causal Analysis and Resolution (CAR) process areas related to the use of process performance models to support “what-if” and sensitivity analysis. The idea is that decision makers will be able to use process performance models to analyze alternative courses of action and alternative improvement ideas. Again, this highlights a capability intended to be exercised within a given project or program execution.
6.The model connects upstream activity with downstream activity. This particular ingredient emphasizes the intent of process performance models to enable decision makers to observe a prediction of the consequences of decisions made earlier in the life cycle or process. Indeed, this ingredient highlights the practical use of process performance models for transitions from phase to phase, hand-offs from one group to another, and so on. This particular ingredient enables the establishment and enforcement of interface agreements between internal groups and/or external groups by providing models that predict the readiness and maturity of an artifact or work product to proceed to the next step. For example, many organizations employ such models to predict defects entering system test while the code is still with the development team. Others use models to predict readiness of design or code to enter an inspection. Still other organizations use models in this fashion to determine if product and software requirements are sufficiently mature and stable to begin intense development.
7.The model enables projects to achieve mid-course corrections to ensure project success. This ingredient highlights a very significant aspect that may be read into the usage of process performance models in CMMI. Specifically, within the QPM process area, process performance models may be used to anticipate undesirable performance with enough lead time to proactively influence the situation toward a successful outcome. Industry experience with this aspect is quite strong, especially in the use of critical parameter management in the Design for Six Sigma (DFSS) community. The notion is that models of critical parameters of the product design foster early insight into issues in products and processes enabling management to take corrective and preventive action. For this reason, organizations employ a collection of process performance models to cover their needs throughout the project life cycle.
