Planview Blog

Ihr Weg zu geschäftlicher Agilität

Produktportfolio-Management, Projektportfoliomanagement, Strategieplanung

Efficient Frontiers and Productivity Rankings in Project Prioritization – Enrich Consulting

Veröffentlicht By Dr. Richard Sonnenblick
Will the real efficient frontier please stand up?

Will the real efficient frontier please stand up?

Each year, we attend many conferences focused on the themes of new product development, innovation, and portfolio management. One of my biggest pet peeves at these conferences arises when speakers discussing project prioritization present a scatter chart with individual projects cumulatively plotted in descending order of bang-for-the-buck (e.g., NPV per dollar of development cost), and refer to it as an ‘efficient frontier’.
(Hint: The faux frontier is the upper graph in the picture)

The graph in question is actually what we prefer to call a productivity ranking. Such a chart features cumulative val

ue (e.g., NPV or eNPV) on the y-axis, and cumulative cost (e.g., cost to launch or cost to the next gate) on the x-axis. The projects are then cumulatively plotted on these axes in descending order of productivity (i.e., value over cost). The projects advance upwards from the origin at a steady pace and then levels out (or trends downwards) as each successively less productive project adds less value and more cost to the overall portfolio.

A true efficient frontier (illustrated by the lower of the two graphs above) is an entirely different beast. Efficient frontiers are built by running multiple optimizations against the set of projects, while varying available resources (development budgets, headcount, or a combination) for each optimization run. The highest value portfolio suggested by the optimizations at each level of resourcing defines the efficient frontier. The result is a curve that may look quite similar to a productivity ranking, but differs in some critical ways.

The productivity ranking is a valid stand-in for the efficient frontier only when some very restrictive conditions hold:

  1. The portfolio does not include multiple development or launch scenarios for each project. Productivity rankings become illegible (even dangerously misleading) when multiple scenarios for each project are included. The optimization specifications used to create an efficient frontier can include multiple exclusive scenarios and will select at most one option for each project under consider while meeting budget constraints.
  2. The portfolio does not include project dependencies. For much the same reason outlined above, a productivity ranking cannot guide the user through nuances of project-technology dependencies, product line dependencies, and even cannibalization. Efficient frontiers built with optimization can easily incorporate all of these factors.
  3. Projects do not vary widely in cost. The productivity ranking is equivalent to the efficient frontier only at precisely the resource levels plotted on the x-axis. In between these resource levels, the efficient frontier usually suggests a higher-value set of projects for the portfolio. This is because the productivity ranking cannot suggest the inclusion of a lower productivity project if your budget cannot accommodate the next highest productivity project, while an efficient frontier built using optimization can.
  4. Projects are all at the same stage in the development life-cycle. This point is especially relevant if your development lifecycle includes more than its fair share of risk. In this situation, projects closer to launch will always look less risky, and hence more productive, than nascent projects under consideration. Thus, a productivity ranking will always favor mature projects over early-stage projects, with potentially disastrous consequences in the medium and long term. When building a true efficient frontier, your optimization rules can contain product launch, revenue, and cash flow targets that will ensure your pipeline is balanced across early and late stage initiatives.
  5. Projects are all in the same division, business unit, or funding bucket. A productivity ranking considers all projects equal across all other dimensions save value. If your project set includes initiatives from multiple business units, a productivity ranking may cause you to unintentionally select projects from only one business unit. The optimization used to build the efficient frontier can accommodate funding guidance for each business unit, and thereby maximize value within each business unit.

I hasten to add that issues 4 and 5 above may also be remedied by simply building separate productivity rankings for each stage and each business unit, and using those assessments to prioritize projects separately by stage and business unit. Taken together with strategic and tactical funding guidance, those prioritizations can then be used to determine the appropriate level of funding, from the bottom-up, for each stage within each business unit. This is a perfectly reasonable way to proceed if you are just beginning your portfolio management journey, and your project set lacks both dependencies and multiple development scenarios.

I’m sure we’ll continue to see some folks refer to productivity rankings as efficient frontiers, but I hope that this has helped clear up the distinctions between the two for many of you. As always, your comments are most welcome! Please send me your thoughts on how we can all improve communication on this vital aspect of portfolio process.

Ähnliche Beiträge

Geschrieben von Dr. Richard Sonnenblick Chief Data Scientist

Dr. Sonnenblick, Chief Data Scientist bei Planview, verfügt über langjährige Erfahrung in der Zusammenarbeit mit einigen der größten Pharma- und Biowissenschaftsunternehmen der Welt. Dank des im Rahmen seiner Arbeit gewonnenen Wissens hat er erfolgreich aufschlussreiche Priorisierungs- und Portfoliobewertungsprozesse, Scoring-Systeme sowie finanzielle Bewertungs- und Prognosemethoden zur Verbesserung von Produktprognosen und Portfolioanalysen entwickelt. Dr. Sonnenblick hat einen Ph.D. und einen Master in Engineering and Public Policy von der Carnegie Mellon University sowie einen Bachelor in Physik von der University of California Santa Cruz.