|
Term
|
Explanation
|
|
A
|
|
|
|
access permissions
|
Specifications that control who can read and/or alter computer files or directories. Web-based project portfolio management (PPM)
tools and other PPM tools intended for multiple users typically use access permissions to ensure that users can view information and take actions that are specific to their defined
roles.
|
|
additive value function
|
Also called additive value model, a value function that may be expressed as a weighted sum of single-attribute value
functions:
V(x1,x2...xN) = w1V1(x1) +
w2V2(x2) ... + wNVN(xN)
As with other value functions, an additive value function assigns a number indicating a decision maker's preference for an outcome, characterized in the formula by the
attributes x1,x2...xN.
With the additive value function, the value assigned is a weighted sum of the decision maker's preferences over the individual
attributes. For example, in the context of project selection, relevant attributes might include the incremental revenue produced, contribution to
organizational learning, and impact on organizational image. The additive value function would compute the measure of project value by weighting and summing individual value measures
indicating how much the decision maker prefers the project's increments to revenue, learning, and image.
Value functions, like utility functions, may be encoded from decision makers by seeking answers to questions regarding
preference for outcomes. The number of required questions and answers increases greatly with the number of attributes used. However, if the value function is additive, it may be
constructed in two steps: (1) assessing the single attribute value functions and (2) assessing swing weights. The single-attribute value functions,
Vi(xi), are, in effect, scaling functions that translate the individual attribute measures into
value units. The weights are called swing weights and may be assessed using the swing weight method.
For the additive value function to apply, several independence assumptions must hold, including preferential
independence, which requires that the decision maker's preferences for achieving any level of performance against any objective does not
depend on the performance achieved against any other objective. Since additivity makes construction of a value function relatively easy, establishing the additive independence of
attributes is one of the most important tasks for multi-attribute utility analysis. Although it may not be possible in a given circumstance to strictly
prove the requisite assumptions apply, there is strong evidence to suggest that so long as the assumptions approximately hold, the additive value model produces results close to what
would be produced without the simplifying assumption.
|
|
aggregation equation
|
An equation used within a decision model to combine attributes, measures of performance, or other numbers assigned to an alternative into a figure of merit intended
to summarize the desirability of that alternative. If, as in the case of project selection, the choice of an option logically depends on the
value of the option, the aggregation equation should include a calculation of value. Thus, the aggregation may include a value function for quantifying value of each alternative).
|
|
agile
|
As used in agile software development, refers to software development practices and methodologies that promote
evolving, timely solutions based on frequent reassessment of requirements and interim deliverables by collaborating, cross-function teams involving customers, designers, and developers.
Similarly, as used in agile project management, refers to project management methodologies that emphasize collaboration, quick delivery time, and
ability to respond to changing requirements. Six sigma and PRINCE2 are examples of a methodologies that
incorporate principles of agile project management.
|
|
algorithm
|
A procedure composed of a sequence of instructions or steps for solving a problem (typically a problem expressed mathematically).
|
|
analysis
|
A systematic approach to building understanding based on identifying and investigating the component parts of a whole and their relationships. The most powerful
methods of analysis involve applying theories and techniques developed within specific fields of expertise, such as those from the natural science, social science, and decision science.
The field of portfolio optimization, for instance, provides powerful analytic methods for prioritizing and optimizing project portfolios.
|
|
analytic hierarchy process (AHP)
|
A popular, theory-based, decision-aiding approach, developed by Thomas Saaty in the early 1970s. Like multi-attribute utility analysis
(MUA) and outranking methods, AHP is a form of multi-criteria analysis. The approach involves
representing a decision problem as a hierarchy. The hierarchy contains the decision goal, the alternatives for achieving the goal, and the criteria for evaluating the alternatives. AHP
provides methods for quantifying the elements of the hierarchy and for evaluating alternative solutions. Some project portfolio management (PPM) tools
use AHP for project prioritization. There are many variations of AHP and how it is applied, making it difficult to provide general
statements about AHP's effectiveness for prioritizing projects.
As originally defined by Saaty, AHP involves asking decision makers to express their preferences over pairs of alternatives with regard to criteria specified in the
hierarchy. For example, "With regard to improving financial performance, do you prefer project A or project B, and by how much?" A nine-point scale is often recommended for specifying
strength of preference (the points on the scale are defined qualitatively, for instance, equal preference, moderate preference, strong preference, etc., see this example). Based on the expressed preferences, a mathematical procedure is used to quantify the hierarchical model for ranking alternatives.
AHP is similar in many ways to MUA. Both are widely used and both reflect a philosophy that alternatives should be selected based on how much they are preferred by
decision makers, not according to some less-relevant goal such as "balance" or "strategic
alignment." The approaches, however, reflect somewhat different philosophies. MUA is regarded as a purely "prescriptive" approach in that it is aimed at identifying the "correct"
decision that a "rational" decision maker would take. AHP, in contrast, is "descriptive" in that it expects errors and inconsistencies in the inputs that people provide and seeks a
solution that best accommodates those errors. The differences in philosophy and methods for eliciting preferences have spurned a lively debate among academics and practitioners over the
pros and cons of the two approaches.
AHP's chief advantage is ease of application. The paired comparison questioning process can be presented pictorially,
and decision makers typically find it easy to make the necessary judgments. AHP produces a decision model that is simpler than that typically
produced with MUA in that AHP does not require the intermediate step of estimating the consequences of projects. AHP is particularly well suited to group decision making since the group
members do not need to be in agreement. Among AHP's outputs are measures of the consistency in the input judgments, which can help participants reduce the inconsistencies expressed in
their comparisons. Also, it is relatively easy for a group to structure a complex decision into the hierarchical structure required by AHP, and the process helps participants to
understand the problem as well as each others' thoughts and opinions.
Saaty has provided an axiomatic theory to support the basic approach to applying AHP. However, the theory has been criticized based on the observation that AHP can
produce "rank reversals," a result wherein the addition of a new alternative can change the ranking of existing alternatives, even though the
new alternative does not influence the costs or benefits of the existing alternatives. For example, it is possible with a project prioritization system based on AHP to have the proposal
of a new project that might be ranked 8th cause a project ranked 4th to move up to 3rd. While such behavior is undesirable and raises questions
regarding the logical defensibility of AHP, supporters have pointed out that rank reversals rarely occur in practice.
AHP's cons include the very large number of comparisons that must be made if there are a large number of alternatives and/or objectives to be considered (that number
can grow exponentially with the size of the decision problem). For this reason, applications of AHP to project prioritization often use a variation of AHP in which the preference
comparisons are expressed for objectives, not for the projects. For example, "Compare the relative importance of the objectives 'improve time to
market' and 'improve financial performance.'" The answers are used to derive a set of weights interpreted to represent the relative importance of the objectives. Projects are then
scored to indicate their contributions to each objective (e.g., no contribution = 0, slight contribution = 0.1,..., excellent contribution = 1). The scores are weighted and added to
obtain an overall measure for ranking projects.
Unfortunately, this common approach to applying AHP may not accurately prioritize projects (it also violates the axioms of Saaty's theory). It is not possible to
obtain meaningful preference comparisons between objectives unless the amounts of improvement are specified (e.g., How can I say how much I prefer the objective "improve financial
performance" without knowing by how much financial performance would be improved?). Also, it is not correct to weight and add performance scores unless the value of achieving a given
level of performance on one measure does not depend on the level of performance achieved on any other measure, a condition known as preferential independence.
An alternative approach to applying AHP overcomes the above problems by using the so-called swing weight method to define
specific amounts of improvements for expressing preferences and by applying tests to ensure that preferential independence holds. In effect, the resulting approach to developing the
model for valuing projects is then consistent with MUA, however, Saaty's AHP's pairwise comparison technique is used to determine the weights.
A final limitation important for applications to project prioritization is AHP's inability to directly address risk. AHP does not allow probabilities to be assigned to
reflect uncertainty over project performance. Decision makers can implicitly factor uncertainty into their pair-wise comparisons, however, this is difficult to do. Although fuzzy logic has been proposed as a means for addressing ambiguity within AHP, nearly all applications of AHP to project evaluation are conducted assuming
most likely scenarios for project performance.
|
|
analytics
|
Term used to describe the methods of analysis used to apply data and mathematical logic to help make decisions. The term usually
refers to the application of more sophisticated forms of analysis. Project portfolio optimization is a common application of analytics.
|
|
application lifecycle management (ALM)
|
The process of managing the phases of the life of a software application, including definition, design, development, testing, deployment, maintenance, and retirement.
Various tools are available to support ALM, addressing such issues as requirements management, architecture, coding, testing, tracking and release management. Project portfolio management tools aimed at IT project portfolios typically include or may be supplied with modules for supporting elements of ALM.
|
|
application portfolio management (APM)
|
The management of an organization's various computer software applications as a portfolio. APM may be viewed as a special case of project
portfolio management (PPM) wherein the "projects" are investments intended to obtain optimal performance from the firm's portfolio of computer software applications. PPM tools aimed
at IT project portfolios may include specialized techniques for managing software applications.
|
|
application program
|
Any self-contained software program that performs a specific function directly for the user. This is in contrast to system software such as computer operating systems
that provide services to support application programs.
|
|
application programming interface (API)
|
An interface provided by a project portfolio management tool or other application
program that defines how that program can access the computer's operating system or other program to request data or services.
|
|
asset management
|
The deliberate, long-term management of an organization's physical assets with the goal of maximizing their contribution to the achievement of the organization's
objectives. Asset management is important for organizations, such as companies in the energy, mining, and oil and gas industries, that must
acquire and utilize expensive assets. Asset management typically involves making decisions about when to create and acquire assets, how to use them, their repair or replacement, and
their ultimate disposal. Project portfolio management applied to asset intensive organizations often focuses on asset management and may be referred to
as such.
|
|
attribute
|
A perceived characteristic of some object. In our context, attributes are measurable characteristics; they are metrics, which may
be objectively or subjectively specified and used to evaluate and compare project alternatives. An attribute, when assigned a number based on some scale, quantifies some relevant characteristic of the project, for example, its cost, how long it would take to implement, or what outcomes it might produce.
The number assigned to an attribute may be a (i.e., a point estimate); that is, a single number intended to represent a best estimate.
Alternatively, if there is uncertainty, a range or probability distribution might be assigned to the attribute. In addition to
being measurable, attributes should be unambiguous, meaning that there is no ambiguity in interpreting the meaning of the attribute or the numbers that are assigned to it. In the
context of multi-attribute utility analysis (MUA) attributes are typically defined to quantify the degree to which an alternative achieves various
decision objectives.
The specification of attributes is a critical step in the construction of a decision model, as the choices made strongly
affect the accuracy, defensibility, practicality, and usefulness of the model. Attributes may involve natural scales, constructed scales, or proxy measures. An
attribute with a natural scale quantifies a project characteristic in commonly used, widely accepted units. For example, cost, expressed in dollars (or euros, rubles, yen, etc.) is a
natural measure for project cost. A constructed scale is a scale that defines different levels for the attribute in terms of descriptions or definitions for each level of
the scale. For example, while it might be difficult to come up with a natural measure for the objective "improve corporate brand image," it might be possible to construct a 1-to-5 scale
consisting of 5 different verbal statements ranging from negative to positive customer perceptions of brand image. A proxy measure is an indirect measure selected because there
is a presumed relationship that exists between it and the relevant project characteristic. For example, if the company concerned about its image participates in a market analysis survey
comparing customer perceptions, its ranking relative to its competitors might serve as a proxy for its brand image.
|
|
audit trail
|
In the context of project portfolio management (non-accounting sense), evidence in the form of records, references, data or documents that enable a user to trace the
path of assumptions made, data changes, decisions, or other past actions that are critical to the results obtained.
|
|
B
|
|
|
|
balance
|
An ill-defined term frequently used in project portfolio management (PPM) literature to suggest that multiple considerations are
relevant for selecting projects. Sometimes, the term is used to argue that the project portfolio ought to contain a mix of projects with different characteristics or objectives, such as
projects that generate near term revenue along with projects likely to generate revenue in the longer term. The term is also frequently used to argue that individual projects ought to
be evaluated against multiple objectives, and that the most desirable projects necessarily strike a balance between positive and negative characteristics, such as risk versus reward.
Many PPM tools provide displays intended to convey the mix or differences in the projects that are contained within a project portfolio, and these can be useful
for developing understanding.
The term balance is most frequently used as an intuitive justification for scoring models. However, because there is no well-defined mathematical measure for
quantifying balance and no available theory for specifying how balance should be optimized, the term is not very useful for serious discussions of portfolio optimization.
|
|
balanced scorecard
|
A popular business process developed in the early 1990s by Robert Kaplan and David Norton for translating an organization's mission and strategy statements into a
quantitative system for measuring organizational performance. Balanced scorecards collect diverse information intended to "balance" the traditional, but narrow, financial view of
performance. The balanced scorecard is an excellent tool for helping managers to understand how the organization is performing and helps translate strategy into action. However,
balanced scorecards are not very useful for prioritizing and choosing projects, and they are often misused in this regard.
According to the balanced-scorecard approach, organizational performance measures should be defined in four areas: (1) finance, (2) customer satisfaction, (3) internal
processes, and (4) innovation and learning for employees. The selected measures are specific to the organization and are chosen to reflect the drivers believed to most important to
understanding success.
As examples, measures of organizational financial performance might include return on investment (ROI), rate of revenue growth,
amount of debt, etc. Customer satisfaction measures might include number of customer complaints, results of customer surveys, average time to process phone calls, etc. Internal process
measures might include fraction of projects delivered on schedule, number of units requiring rework, process yield rates, etc. Learning measures might include number of employee hours
spent in training, numbers of employee suggestions submitted, etc. Balanced scorecard measures can be backward-looking, to monitor how the organization is doing, or forward-looking, to
assess the future impacts of alternative courses of actions. Assessments against the measures are arrayed on pages or displays referred to as "scorecards." Target levels of performance may be assigned to the measures. Balanced scorecards are being used in a broad range of activities, from
product planning to incentive compensation, and by federal, state, and local governments.
The major weakness of balanced scorecards is that the approach does not provide a basis for trading off performance on different measures. In other words, if an
organization improves performance on one measure without degrading performance on any other measure, that is a good thing. However, if making a change intended to boost performance in a
given area (e.g., customer satisfaction) threatens performance in some other area (e.g., finance), a traditional balanced scorecard cannot indicate whether that change is desirable or
should be made.
In an attempt to address the above weakness, balanced scorecards have sometimes been expanded to include a means for aggregating individual performance measures into a
quantity meant to represent the overall performance of the organization. For example, most commercially available tools for project portfolio management allow users to define equations
for combining performance measures—A scorecard is defined for assessing projects, and the various measures on the scorecard are mathematically combined to provide an indicator
intended to represent the relative desirability of the project. Typically, the form of the equation is weight-and-add (sometimes, the performance measures are merely averaged, which,
presumably, implies a desire to weight the measures equally).
Choosing projects so as to maximize a weighted sum of scorecard measures is almost always incorrect (see preferential independence). Yet, some try to justify this approach by arguing that organizations should strive for balance in performance across various areas and measures. However, maximizing a weighted average does not necessarily lead to balance (if balance means
including projects that address all measures). In any case, the goal of project selection is to choose projects that create the most value, not balance (whatever balance means).
A project that has a high weighted-average performance score may or may not be a high-value project. How well the weighted-average score relates to value depends on,
among other things, how the measures are defined, the number of measures within each area and the degree to which they overlap or "double count," the organization's current level of
performance, and the basic objectives of the organization. It is possible to define performance measures that can be aggregated into an overall measure of project value, but this
requires a different process for defining performance measures than that used in the balanced scorecard approach (see multi-attribute utility analysis
(MUA).
|
|
