|
Term
|
Explanation
|
|
P
|
|
|
|
paired comparison
|
Also called pair wise comparison, a process for simplifying decisions or judgments that involves comparing options or entities in pairs and judging which element
of the pair is preferred or has the greater amount of some other measurable property. Among other applications, paired comparison is used as a method for assessing preferences, assigning
probabilities, voting, and ranking decision options.
Paired comparison is sometimes used as a method for prioritizing projects. Each project option is compared against each other option, and a score is assigned to each option
under each comparison to indicate the degree of superiority or inferiority. The pairwise comparisons are then combined in some way (e.g., summed) to obtain an overall figure of merit
for each project.
|
|
payback period
|
The amount of time it takes for the cumulative cash flows from a project to equal the initial investment. An investment will have
paid for itself in the year, or month, where the cumulative cash flow first becomes positive. Payback period is a popular metric for evaluating
projects because of its simplicity, emphasis on liquidity, and obvious responsiveness to external financing pressures. However, because payback period does not provide an adequate
measure of project or portfolio value, it should not be used as a metric for prioritizing projects.
A major limitation of the payback period method is that it ignores cash flows after the payback period. For example, a small project may have a break-even point at six
month from the start of the project. A larger project that costs twice as much may not break-even until 12 months after the start of the project. Based on the analysis of the payback
periods, the company may choose to go with the smaller project. The analysis would ignore the fact that after two years, the larger project would have produced three times the dollar
savings as the smaller project. Also, as no discounting is involved, the payback period overlooks the time value of money (cost of capital).
|
|
perception biases
|
A category of cognitive biases (distortions in judgments) characterized by faulty perceptions. Examples of perception biases include overconfidence, anchoring, base-rate and bounded awareness.
|
|
performance measure
|
A metric used in a decision model to evaluate and compare alternatives, such as alternative projects, based on the outcome or outcomes that would result from those alternatives. If a project outcome is uncertain, a point estimate may be chosen or a range or probability distribution may be assigned to a
performance measure to represent that uncertainty. Using performance measures to evaluate and prioritize projects is generally regarded as best practice. However, as explained below,
most project portfolio management (PPM) tools make only limited use of performance measures.
The term performance measure is similar to but distinct from the related terms metric, criterion, and attribute, all of which appear frequently in PPM literature:
- A metric is any measure used to quantify some aspect of business performance, not necessarily one with characteristics suitable for evaluating
projects or supporting project selection.
- The term criterion, on the other hand, refers to anything that might be used to evaluate an alternative, and is typically employed in
situations where multiple criteria are relevant (i.e., where multi-criteria analysis is required). A challenge for using criteria
for project prioritization is that unless the criteria are defined so as to meet certain requirements, there is no general,
logically-defensible way to aggregate or roll up criteria into a combined measure of how good or attractive each alternative is (e.g., it can be shown, in general, that it is
logically incorrect to weight and add scores assigned to criteria).
- An attribute is a special type of criterion that meets the requirements for multi-attribute utility analysis
(MUA), an approach for creating a decision model that includes techniques for determining a logically sound aggregation equation
for combining the attributes. Note that whereas the term metric refers to a measurable business outcome, the term attribute need not be a decision outcome directly relevant to
measuring business performance. For example, the number of executives who support a project might be defined as an attribute, but by our definition it is not a performance
measure.
As used here, the term performance measure refers to a special kind attribute that, in addition to being measurable, operational, and understandable (requirements for
a well-defined attribute), is also an observable consequence or outcome of the decision. Well defined performance measures pass the clairvoyant
test.
There are important advantages to using performance measures (rather than other kinds of criteria and metrics) to evaluate and prioritize projects. First, and most
obvious, the purpose of projects is to create outcomes desired by the organization. Thus, evaluating projects based on their likely outcomes is the most direct approach. Also, if
projects are prioritized based on forecasts of impacts on business performance, then the organization has the opportunity to compare forecasted performance with the performance that
actually occurs. Importantly, this gives the organization the opportunity to learn and therefore improve its decision model and the processes it uses to make forecasts. Furthermore,
using impacts on performance to evaluate proposed projects reduces gaming. If the organization compares forecasts with actuals, individuals who
might consider biasing estimates risk having their biases exposed. Interestingly, experience shows that gaming is less likely with performance measures even if the organization makes no
such comparisons. Perhaps, knowledge that with performance measures the organization could make comparisons encourages project proponents to take more care when generating
forecasts.
Given the advantages of using performance measures to evaluate projects, it might seem surprising that most PPM tools do not take this approach. Instead, the most
common approach selected by PPM tool developers is to allow users to score projects (e.g., on a 1-to-10 scale) against criteria defined by the user. The tool then ranks the projects
based on the weighted and summed scores. As noted above, weighting and adding criteria scores is always incorrect. Although most tool developers realize this is wrong, there are several
reasons for taking this course: (1) it is simple and easy to implement, (2) the approach is generic and allows the same tool to be marketed to virtually any project-based organization
in any industry, and (3) most customers don't realize the weight and add approach gives incorrect recommendations. Oftentimes, the tool vendor adopts a pseudo-scientific term, such as
strategic alignment, to describe the weight-and-add approach, which helps lend a sense of false credibility to the flawed
logic.
Because different types of projects produce different consequences, and because the desired project consequences depend on the type of organization and its mission,
PPM tools that recommend projects based on performance measures are necessarily industry specific. Thus, in order to obtain accurate project portfolio recommendations, it is generally
necessary for an organization to seek a PPM tool designed for an organization of its type and for the types of projects to be included within its project portfolios.
|
|
PERT chart
|
A diagram that graphically displays a type of network model often used to support the planning and analysis of projects composed
of many interdependent tasks. PERT stands for Program Evaluation and Review Technique. PERT charts were first developed in the 1950s by the Navy to help manage very large, multi-step,
non-routine projects having a high degree of inter-task dependency.
In a PERT chart, tasks are represented by nodes and task dependencies are represented by lines connecting the nodes.
A simple PERT chart
A primary use of PERT charts is to compute the minimum time required to complete the project, where the model allows for uncertainty in the time necessary to complete
each task. Project management tools typically support PERT charts and some portfolio management tools provide this capability as well.
|
|
platform project
|
A project that creates knowledge or capability (i.e., a "platform") for delivering other projects, such as a new generation of product or service offerings. An example would be a
project for an automobile company to create its first electric vehicle. Platform projects represent significant departures from existing offerings and are often more costly and/or
risky, but may generate strategic value for the organization.
|
|
PMBOK®
|
Refers to a project management guide provided by the Project Management Institute, a non-profit professional organization. PMBOK stands for Project Management Body
of Knowledge. The PMBOK guide specifies widely recognized and accepted standards for project management information and practice.
|
|
PMI analysis
|
A simple decision making aid wherein the pros, cons, and interesting implications of a project or other proposed action are listed and evaluated. PMI stands for plus,
minus, and interesting. Positive and negative scores are assigned subjectively and the results added. A strongly positive total is interpreted as suggesting that the action should be
taken and a strongly negative score that it should be avoided.
|
|
point estimate
|
A single numerical value selected to summarize what is, in reality, an uncertainty that could take on any value within some range of possibilities. Uncertainties are
more accurately described by probability distributions.
|
|
portfolio balancing
|
A term often used in project portfolio management (PPM) to describe the last step in the process of identifying an optimal project
portfolio. Tools for PPM facilitate this step by providing portfolio mappings, graphical displays that show the mix, or "balance,"
of projects across various dimensions (e.g., risk versus return, long-duration versus short duration, maintenance versus asset enhancement, incremental
projects versus breakthrough projects, etc.). Tool vendors assert that the displays allow instances of imbalance to be identified and help suggest project
substitutions that improve the portfolio.
It's a bit more complicated than that. While PPM literature frequently advises portfolio balancing, there is very
little real advice about how to assess portfolio balance or what changes should be made to the distribution of projects to improve balance.
As I argue in my papers, the goal of PPM is to select the set of projects, subject to applicable constraints, that
creates the greatest possible value, where value is defined as the worth to the organization of the consequences that would be produced by conducting all of the projects in the portfolio,
and with portfolio value adjusted to account for portfolio risk in accordance with the organization's risk tolerance. Thus, the optimal project mix is that which maximizes risk-adjusted, portfolio
value.
Nearly all PPM tools include some model or logic for prioritizing projects or recommending project portfolios. For most tools, the model is very simple
(e.g., a weighted scoring model). Many tools rank projects based on scoring strategic alignment,
an approach that has little to do with measuring project value. Only a very few PPM tools utilize sophisticated value models and accurate
optimization algorithms. The reason for this is that defensible value models need to be organization and project-type specific,
and vendors can't make as much money selling a tool that only fits a narrowly defined market.
Regardless of the tool used, because even the best models are not perfect, the project portfolio
that is recommended by a tool can never be assumed to be a perfectly accurate identification of the value-maximizing project portfolio. Thus, it is always necessary to consider
how the recommendations produced by a tool (or obtained by any other mechanism) ought to be adjusted and improved. One could call this subjective step
"balancing the portfolio," but it might also be called accounting for considerations that are not well addressed by the model.
For example, if projects involve risk, but the model does not adequately account for the impact of diversification and risk tolerance when computing
portfolio value, then it will be particularly important to check whether the recommended portfolio presents an acceptable level of risk and whether it
reflects a bias toward either risky or safe projects. If the
application includes multi-year projects or projects that generate benefits over time, but the model does not capture time preference, or if it recognizes constraints on
only current year and not out-year spending, then it will be
important to check whether the portfolio provides a reasonable balance between immediate and long term costs and benefits, and to check whether portfolio cashflows are sustainable (e.g., will near term cash flows
be sufficient to support investments necessary to assure returns over the long term?). If
the different types of projects require different people resources (e.g., research projects versus development projects), but the model doesn't account for constraints
on the specific resources needed by each type of project, then it will be important to check whether the recommended portfolio balances demand for the available types of constrained resources.
If there are dependencies among projects, but the model merely ranks projects without regard to the dependencies, then changes may similarly be needed to
better "balance" the portfolio so as to take such dependencies into account.
The point is that it is always necessary to examine a candidate portfolio carefully from various perspectives for the purpose of identifying changes that would
increase portfolio value. Portfolio balancing may be as good a term as any for this step. In any case, a PPM tool should provide graphical outputs that assist
portfolio balancing. However, because absent a better model there is no way to measure whether a change to "balance" actually increases portfolio value, it is
best to start with a quality model, thereby reducing the extent to which subjective judgment unsupported by analysis is needed to translate the "bad
recommendations" made by a poor tool into choices that actually create more value for the organization.
|
|
Previous