This Guest paper was submitted for publication and is copyright to Gary J. Summers © 2009
Published October 2009

Introduction | The Bayes' Law PPM Model | Proposals and Selection
New PPM Metrics | Management | Improving Your PPM Situation | Conclusion

The Bayes' Law PPM Model

Can you use the new model?

Because PPM results represent a non-random selection of proposals, you must analyze PPM results with techniques that match your method of selecting projects. You can use the new PPM model if you select proposals by (1) evaluating proposals, (2) prioritizing proposals, (3) selecting down the ranking and (4) making changes for various reasons.[2] The changes must not be too extensive, and you must evaluate projects with an interval or ratio scale. For example, you can use financial metrics, decision analysis models, the analytic hierarchy process (AHP) or scoring models. You will get the best results if you classify proposals into strategic buckets and then prioritize and select from each bucket. If you use optimization, the model may work for you as well.

While some metrics require less data, measuring the quality of your project prioritization requires data from at least fifty project proposals and thirty-five funded projects. These requirements are not as extreme as they appear. The new model analyzes PPM results, so you can use past PPM data. Suppose data from the past two years is relevant. You can use the model if, on average, a strategic bucket evaluates twenty-five proposals and funds eighteen projects annually. If data from the past three years is relevant, a strategic bucket must evaluate seventeen proposals and fund twelve projects each year. Depending upon the outcome of my research, you may be able to pool data by combining data from several strategic buckets.

Bayes' Law and PPM

If you evaluate PPM results, you will evaluate a funded project (at least) twice: once before project selection and again after the project is completed. When a project is evaluated before selection, I refer to it as a proposal; when it is evaluated after selection, I refer to it as a completed project. I use these terms even though some proposals and "completed" projects are currently ongoing.

The most important feedback metrics require a simple evaluation of completed projects. You must classify them into two categories: Good and Bad. You can define Good and Bad in any way that suits your business. The categories apply to proposals as well. A Good proposal is one that, if it is funded, will produce a Good completed project. A Bad proposal is one that, if it is funded, will produce a Bad completed project.

With these definitions in mind, the odds version of Bayes' law governs PPM as follows:



  •  P-proposals is the fraction of proposals that are Good proposals.
  •  P-results is the fraction of completed projects that are Good projects.
  •  QPS is the quality of project selection

Obviously, high values of P-results boost financial performance. When P-results is high you have more Good projects creating value and less Bad projects wasting resources. The next sections reveal how P-proposals and QPS affect P-results.

Introduction  Introduction

2. Editor's Note: Note the sequence advocated here by the author, namely: "(1) evaluating proposals, (2) prioritizing proposals, (3) selecting down the ranking and (4) making changes for various reasons." By way of comparison, the Project Management Institute's Standard for Portfolio Management calls for: "Identify; Categorize; Evaluate; Select; Prioritize; Balance ..." (Chapter 3, Section 3.3, 2008). If you Select then Prioritize rather than Prioritize then Select, as required by the author, your results may be impacted to some degree.
Home | Issacons | PM Glossary | Papers & Books | Max's Musings
Guest Articles | Contact Info | Search My Site | Site Map | Top of Page