Project Portfolio Management - Maximizing Value With Real Optimization

Suppose you have 30 projects that would cost a total of $40 million and require 30 full-time employees, but you only have a budget of $20 million and 22 full-time employees.

And what if you want a minimum ROI of 15%?

And what if you want to control the average portfolio risk?

In a 30-project portfolio, there are over 1 billion possible project combinations. In a 40-project portfolio, there are over 1 trillion!

So how do you pick the set that will deliver the most value and not exceed your constraints?

Real optimization allows you to find a set of projects that yields the maximum portfolio value when you have cost, resource or other constraints.

Some Project Portfolio Management (PPM) applications "optimize" against cost by simply starting at the top of the project list and stopping at the point where the money runs out. That's not optimization at all.

Besides the fact that this approach treats project cost as the only constraint and ignores other constraints such as resources and time, what if there were several projects below the one where you stopped that would together deliver more value than one of the selected projects and also cost less?

You'd miss that value entirely.

Furthermore, this approach does not consider the portfolio's "efficient frontier." The efficient frontier represents the maximum output (such as profit) that can be achieved for a given input (such as cost). To find the efficient frontier for portfolio, you would take the output value for each project and divide it by it's corresponding input value and then order the projects from highest to lowest. For example, ordering a portfolio by profit/cost would rank the projects by the most units of profit per unit of cost.

Using efficient frontiers can show you where the greatest "bang for your buck" is. When you've scored your projects by value, you can get an efficient frontier of the value-score/cost to see portfolios that will give you the most value per unit of cost.

So we can see that simplistic approaches to portfolio optimization won't work for finding optimal value-maximizing project sets against multiple constraints.

Unfortunately, most Project Portfolio Management systems do not offer an integrated optimizer that can perform rigorous optimizations against multiple constraints.

A Project Portfolio Management tool that ranks projects without the capability of performing true optimizations is of limited usefulness. Consider systems that have built-in optimizers integrated with a sound prioritizing system. Remember, an optimized portfolio can only be as good as the ranking system that was used to rank the projects.

What should you look for in an optimizer?

There are basically three types of optimizers:

  • "Brute force" optimizers check every possible combination of projects to find an optimal one. These work well for small portfolios, but rapidly become far too time-consuming for larger portfolios.
  • Stochastic optimizers use techniques such as "linear programming" and "integer programming" to find optimal portfolios. These work well for larger portfolios, but can also become time prohibitive for very large portfolios.
  • Heuristic optimizers use algorithms developed from artificial intelligence research such as "genetic" or "evolutionary" algorithms to find optimal and near optimal portfolios. A well-designed heuristic optimizer will work well for large portfolios, and will usually discover a near-optimal solution quickly.

Look for optimizers that have a user-friendly interface and allow simple menu-driven setups. Unless you have a strong propensity for doing mathematics, avoid optimizers that require you to set up your constraints in the language of linear programming.

Look for flexibility in the kind of constraints that you can use beyond just "not greater than" and "not less than" constraints. "Average not less than" and "average not greater than" can be useful for obtaining overall portfolio outcomes related to time and risk and reward constraints.

Also, be sure that you can optimize to different attributes and criteria including on efficient frontiers.

Make sure that your optimizers allows you to "force-in" and "force-out" projects into the portfolio.

Finally, make sure that the application has the capability of easily comparing different portfolios optimized with different constraints.

In summary, "Brute force" optimizers are not suitable for large portfolios. Linear programming optimizers can find optimal solutions but may be limited in the types of constraints available. Heuristic optimizers can find optimal and near-optimal portfolios quickly with more types of constraints. Look for a flexible user-friendly interface and the ability to easily compare portfolios with different optimization constraints

If you're looking for a Project Portfolio Management solution with state-of-the-art but easy to use optimizers for robust and defensible project portfolios, then click here to take a look at Optsee®, our project portfolio management tool. We have "cracked the code" of project selection by making it easy for ordinary business people to apply state-of-the art business analytics to project prioritization and portfolio optimization for results that are both understandable and defensible.

Next: Project Portfolio Management - Getting the Best Organizational Fit

Click Here To Try Optsee® Today!