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


Managing Resources

Recall from Figures 4 and 5 that the office laser printer bucket had better proposals and prioritization than the professional printing bucket. Subsequently, Figures 6 and 7 showed how this advantage produced a better financial opportunity with laser printers. Based on financial performance, one should invest more (larger bucket) in laser printers and less (smaller bucket ) in professional printing. Of course, resources also limit investments in opportunities. By comparing resource constraints to the curves in Figures 6 and 7, you can see if a bucket needs more resources or if some of its resources should be invested elsewhere.

To allocate resources well, you must also consider the impact of resource allocation on individual projects. Funding too many projects will spread resources thinly and cause your projects to suffer. You will turn Good projects into Bad projects and decrease P-proposals. (I am grateful to Max Wideman, project management expert and author, for proposing this mechanism.)

The new model can help you solve this problem, but only if your proposal evaluations consider the adequacy of your resource commitments. Your financial metrics and decision analysis models should consider multiple resource levels, and your scoring models should include resource levels as an attribute. If you evaluate proposals this way, a proposal's score (evaluation) will decrease when your resource allocation plan reduces its resources.

With this type of evaluation, Figures 8 and 9 will help you assess the impact of your resource allocation on your projects. As you change the amount of resources dedicated to each project, each project's score (evaluation) changes. Figure 8 then shows the impact on each project's chance of success. Derived from your PPM results, Figure 8 provides the probability that a proposal is a Good one, based on its score. Figure 9 shows the probability of success for each project and for P-proposals. The light bars indicate proposals that have a low probability of success because the current resource allocation plan provides them with inadequate resources. (By estimating the likelihood of success for each project, Figure 8 can help you assess and manage portfolio risk as well.)

Figure 8: Probability of success as a function of project score
Figure 8: Probability of success as a function of project score
Figure 9: Probability of success for each project and P-proposals
Figure 9: Probability of success for each project and P-proposals

Because resource allocation affects each proposal's probability of success, resource allocation affects some of the aforementioned metrics. A PPM dashboard should adjust the metrics so you can see the full impact of your resource allocation plans. Hopefully, this feedback will eliminate any tendency to fund too many projects.

Managing Phase-Gate Systems

The Bayes' law model shows how Phase-Gate systems behave. Consider the Phase-Gate system in Figure 10. Selecting more proposals at Gate 1 increases pipeline throughput, but it lowers QPS-1. A lower QPS-1 reduces P-development, so the pipeline has more downstream failures. Conversely, selecting fewer proposals at Gate 1 reduces throughput, but it raises QPS-1. A higher QPS-1 raises P-development, so the pipeline has fewer downstream failures. Downstream failures are more expensive than upstream failures, thus:

Pipeline throughput and ROI are inversely related.
Greater throughput reduces ROI.

Additionally, when P-market is small, executives are concerned with the consistency of output. Greater throughput lowers P-market, and from the theory of geometric distributions, a smaller P-market makes a pipeline more volatile. Increasing throughput intensifies booms and busts.

Figure 10: The behavior of Phase-Gate systems
Figure 10: The behavior of Phase-Gate systems

These relationships identify the following problems with current pipeline management:

  • Selecting projects to maximize NPV at each gate manages each gate independently. This practice ignores the relationship between upstream project selection and downstream results.
  • Some textbooks and models assume that the gates' attrition rates are independent of each other. This assumption is wrong.
  • Pushing more projects into the front-end increases expensive downstream attrition while raising pipeline volatility.
  • Plans calling for high throughput with high ROI are infeasible and will fail.

The Bayes' law model suggests a new approach to pipeline management. With PPM results, you estimate P-proposals, QPS-1, and if sufficient data exists, QPS-2. You then select the level of throughput that is best for your company. To implement your plan, raise Gate 1's cutoff value (or hurdle rate) high enough to reduce Gate 2's attrition to a desired level. Conversely, if Gate 2's attrition rate is too high, fix the problem by raising Gate 1's cutoff value. By measuring results downstream and fixing problems upstream, you coordinate project selection throughout your pipeline.

New PPM Metrics  New PPM Metrics

Home | Issacons | PM Glossary | Papers & Books | Max's Musings
Guest Articles | Contact Info | Search My Site | Site Map | Top of Page