Professional Documents
Culture Documents
Four Quadrant Expectation Management: All of DW
Four Quadrant Expectation Management: All of DW
All of dW
Expectation traceability
from The Rational Edge: This article defines a software project management technique called "four quadrant Applying four quadrant expectation management
expectation management" and presents an example of expectation modeling. Included are an explanation of Example: Creating an expectation model with IBM Rational RequisitePro
how to implement the technique using IBM Rational RequisitePro and a brief look at how the technique can Applying expectation management across the organization
Notes
Many software projects fail. Considering all the About the author
To a large extent, people declare that a project has either succeeded or failed based on whether it met their expectations. Few projects fail in an absolute sense --
they simply fail to meet individual expectations.
What are expectations? In a software project, requirements are explicitly stated expectations that users place on the project team. Ideally, the team and the users
should interact iteratively to create and solidify requirements, but typically requirements originate as user needs or features that the project team captures in a
requirements gathering process. A budget, for example, is an expectation that sponsors place on the project manager. A timeline is an expectation the project board
places on the project manager. Other expectations are not explicitly stated. For example, developers might expect to use their same old comfortable but ineffective
development process. An IT director who secures a project by underplaying the resource and time requirements might expect the project to make him look good and
gain him a promotion. Such implicit expectations tend to shape the explicitly stated ones (just as our sub-conscious mind shapes our conscious expression). It is
important that the organizational players who might be affected by these implicit expectations try to discover them; they may actually carry more weight than the
stated expectations and must be managed to ensure project success.
with tools -- in much the same fashion as the team manages requirements. To do this correctly, the project needs a part- / full-time expectation analyst throughout its
lifecycle.
This person would begin by defining formal expectation management (similar to requirements management in RUP), using tools, such as IBM® Rational®
RequisitePro.® He or she would classify explicit expectations as "stated" and implicit expectations as "perceived." These can be further classified as either "internal"
to a domain or belonging to "external" parties.
Figure 1 shows the expectations that a project team needs to fulfill, including requirements and many others. Communication flows from internal to external, and
clarity improves as an expectation goes from perceived to stated. As this figure shows, if you manage requirements alone, you take into account only external-stated
expectations and completely miss out on the other three types: external-perceived, internal-stated, and internal-perceived.
One day the customer's CEO meets the company's CEO while playing golf.
Customer CEO (to company CEO): The market is changing fast. We wish our suppliers would realign their products with these changes!
The company CEO returns to work and summons the company vice presidents of business and IT, respectively.
Company CEO (to company business VP): Our customer indicated that we need to realign our product with their needs. Why don't you work
with the customer and Jim here, our VP of IT, to make that happen?
Four quadrant expectation management
Company CEO (to company IT VP): I think you need to revamp our IT strategy. Why don't you start an IT strategy project and work with Susan
here, our VP of business, on the objectives?
Eventually, the above communication results in a project to migrate the product to IBM® WebSphere.® As Figure 2 shows, an expectation analyst creating a model
for this project would treat the company as the domain. Each subject's expectations can be classified into a quadrant: external-stated, external-perceived, internal-
stated, or internal-perceived. For example, the customer CEO's "realign product" message is an external-perceived expectation. It comes from someone outside the
company, and it is perceived by the company CEO as an expectation. The measure of confidence, or sureness factor, for this expectation is 75 percent. By
implication, 25 percent of the expectation is open to interpretation. This percentage is a subjective call, based on how confident the expectation analyst feels that he
or she understands the expectation.
The results of expectations in the model shown in Figure 2 become elements of project planning, scoping, and management (e.g., budget and schedule in the
software development plan or features in the software requirements specification.) These elements also have associated sureness factors. For example, although
the project manager has an internal-perceived expectation that the budget will be $1.2 million, based on conversations with the VP of IT, s/he tells the expectation
analyst to stay within the internal-stated budget of $1 million that the VP puts in writing.
The system analyst advises the expectation analyst to assign the "share data" feature a sureness factor of 50 percent, based on an internal-perceived expectation
held by the customer marketing analyst. The system analyst is 75 percent sure that the marketing analyst actually wants to join the customer company, and therefore
might be promoting this feature for her future benefit. In such instances, the system analyst might decide to validate only those features whose sureness factor is at
or above a specific limit -- say 80 percent -- as firm requirements.
This example shows how important it is for the expectation analyst to consult with other team members in order to discover "hidden" expectations and accurately
estimate the sureness factors.
Now that we have explained the basic benefits of expectation management and the elements of an expectation model, let's look more closely at the four quadrant
expectation management technique. We'll begin with a few definitions.
Expectation
Expectation management
A formal process to continuously capture, document, and maintain the content, dependencies, and sureness of the expectations for persons participating in an
interaction, and to apply the information to make the interaction successful.
Expectation objective
The objective for an interaction that is under expectation management. For example, the interaction might be an important meeting whose objective is to convince a
CEO to release a project budget. The meeting objective is also the expectation objective, which can be used to capture corresponding expectations.
Expectation domain
A set of participants sharing a common expectation objective for the interaction under expectation management.
Expectation model
A complete description of expectations for a system that are held by interacting persons at a given time.
Expectation manager
Expectation subject
Expectation object
Expectation result
A real number (>=0, <=1) denoting the level of sureness in an expectation or expectation result, 1 being the maximum.1
The following four definitions relate to the scope and clarity of expectations:
External expectation
Internal expectation
Stated expectation
Perceived expectation
Expectation quadrants
Figure 3 shows how the expectation quadrants work. The horizontal (x) axis demarcates for clarity: Expectations above the x-axis are stated; those below it are
perceived. The vertical (y) axis demarcates for scope: Expectations to the right of the y-axis are external; those to its left are internal.
Expectation traceability
In addition to classifying expectations according to the quadrants defined above, it is desirable to establish traceability for expectations. Expectations generate other
expectations, including project budgets, timelines, and requirements, as we have seen. They also generate meeting objectives, management objectives, risks, and
other results that shape much of our work and its outcomes. This "expectation chain" gives rise to the concept of expectation traceability, essentially a broader
application of requirements traceability as defined in RUP that applies beyond requirements and includes sureness factors. The clarity level for an expectation
(stated versus perceived) results in a sureness factor between 0 and 1, depending upon heuristics as well as preceding expectations -- their own sureness factors
and how they occur, whether in a sequence (AND) or all together (OR). In addition, we can assign a sureness factor to expectation results based on the sureness
factors for preceding expectations.
Decisions about sureness factor values are influenced by traceability type. We will look at two types below: AND traceability and OR traceability.
AND traceability
AND traceability is possible when an expectation (or expectation result) traces out of one other expectation, which in turn traces out of another expectation, and so
on, as shown in Figure 5.
● Expectation or result tracing out of the AND expectations = less than or equal to the smallest sureness factor from the AND constituents.
Four quadrant expectation management
OR traceability
OR traceability is possible when an expectation (or expectation result) traces out of more than one other expectation, as shown in Figure 6.
● Expectation or result tracing out of the OR expectations = greater than or equal to the largest sureness factor from the OR constituents.
Define an objective
It is helpful to start with defining an objective for your management strategy, so that you will understand what the domain is and who the subjects are.
For example, your objective for a software project might be "to manage expectations of the customer, CEO, business, and IT teams in relation to scope, budget, and
schedule for project X. Or, if you're holding a meeting to convince a CEO to release budget for a project, the objective might be to capture expectations of the CEO
and his / her direct reports with respect to budget allocations. You might also want to capture their expectations regarding major application projects, including the
project for which you're seeking a budget.
Once you've defined an objective, you can define domain boundaries. Defining the domain will enable you to better understand the project scope -- and whether an
expectation is external or internal. Any people who work outside of the domain boundaries should be considered external.
Next, you need to identify those who will play key roles in satisfying your objective, both within and outside the domain. These are the persons whose expectations
you want to manage.
Below are steps for setting up Rational RequisitePro for expectation management (assumes functional knowledge of the tool).
Create a new requirement type called expectation with the tag EXP, and then create the attributes shown in Table 1. Note that these include the
parameters associated with our expectation management quadrants.
Table 1: Attributes for the expectation requirement type in IBM Rational RequisitePro
Domain Single value Name of the expectation domain and names of elements outside it -- for example, company (domain) and customer (external element).
Create a new requirement type called Software Development Plan with the tag SDP. Then create the following attribute (in addition to others you might
need):
❍ Sureness factor: Real value
3. Add the attribute sureness factor to the type expectation and other dependent requirements
Add the attribute sureness factor (with a real assigned value) to other requirement types that could depend on expectations (e.g., stakeholder requests,
features, etc.).
Create a package named Expectations. Within this package, create a package for each subject.
5. Create expectations
Create the attribute, traceability matrix, and traceability tree views as necessary. When creating traceability trees, be sure to include "traced out of" for all
requirements that depend on expectations or their dependents.
7. Create traceabilities
Create traceabilities between the requirements of type EXP (the Expectations-to-Expectations traceability matrix), and from other requirement types, such
as FEAT, SDP, and so forth, to EXP. Make sure you check "Show Indirect in View->Properties" when displaying the Expectations-to-Expectations
traceability matrix.
Capture expectations
Once you have set up IBM Rational RequisitePro to accommodate expectations, you can begin their capture. Interview each subject, discussing their expectations
around the project objective. Then, classify these expectations by quadrants and enter them into RequisitePro in appropriate packages, as we will see in an example
below. You should continuously monitor and update the expectations, including their descriptions, sureness factors, and traceabilities. You should also update any
plans affected by these changes.
Expectation management can also improve intra- and inter-team communication by promoting a better understanding of stakeholders' stated and unstated
expectations. It can capture hidden expectations that render useful information about a project's success criteria and risks.
Business interactions unrelated to software project management can also benefit from expectation management, as we shall see below.
Imagine that you wanted to sell process improvement services to a company. You could draw the organization chart in Figure 11; the emphasized lines show
reporting relationships among the CEO, vice president of IT, vice president of business, and project manager. Through earlier interactions with company officers, you
already know the following:
● The CEO definitely approved a project budget of $$$ requested by the VP of business.
● You are 80 percent sure that the project manager expects to be nominated as lead project manager by the VP of business.
● You are about 60 percent sure that the project manager is hoping to be appointed VP of IT when an imminent reorganization takes place.
Under the circumstances, if you have been talking to the VP of IT regarding a possible process improvement service offering, your strategy needs to include and
Four quadrant expectation management
align with those of the project manager and VP of business. As Figure 11 shows, reporting structures do not necessarily mirror expectation structures.
Note that, for any of the applications we have discussed, you can reduce expectation management to a quick, informal analysis if appropriate. For example, before a
Four quadrant expectation management
meeting, it may be enough to simply draw a quick expectation model on a whiteboard in order to understand the expectations of those coming to the meeting.
Conclusion
All people have expectations that drive the way they interact, whether they are at a family gathering, attending a meeting, working on a project, managing a business,
or even leading a country. Understanding these expectations and responding to them is an art for managers; expectation management is the corresponding
discipline, and four quadrant expectation management is a technique within that discipline. It is a useful technique that applies not only to software development
projects, but also to sales planning, contract negotiations, project management, meeting management, and any other area in which human beings must collaborate
effectively to achieve a shared result.
Notes
1Based on the concept of Certainty Factors (see discussion on the expert system MYCIN: E. H. Shortliffe. Computer-Based Medical Consultations: MYCIN. Elsevier,
New York, NY, 1976).
Strongly disagree (1) Disagree (2) Neutral (3) Agree (4) Strongly agree (5)
Comments?
Submit feedback
developerWorks > Rational >