AGILE PLANNING FOR

SOFTWARE PRODUCTS

COURSE NOTES

Agile Planning for Software Products | 1

Table of Contents
Course Overview ............................................................................................................... 4
Introduction ................................................................................................................................4
Module 1: Introduction to Planning ................................................................................ 6
Introduction to Planning ...........................................................................................................6
Important Terms ................................................................................................................................. 7
Previous Concepts ............................................................................................................................. 7
Uncertainty Space ......................................................................................................................8
Navigating Uncertainty Space Diagrams ........................................................................................ 9
A Final Note on Uncertainty Spaces ............................................................................................. 11
Work Breakdown Structure (WBS)......................................................................................... 12
Example Work Breakdown Structure ............................................................................................ 13
Creating a Work Breakdown Structure ......................................................................................... 15
Uses of Work Breakdown Structures ............................................................................................ 15
Estimates, Targets, and Commitments ................................................................................. 16
Estimates ........................................................................................................................................... 16
Targets ............................................................................................................................................... 17
Commitments .................................................................................................................................... 17
Estimate, Target, and Commitment Example .............................................................................. 17
Module 2: Estimation ...................................................................................................... 19
Story Points .............................................................................................................................. 19
How To Use Story Points ................................................................................................................ 19
Advantages of Story Points ............................................................................................................ 21
Limitations of Story Points .............................................................................................................. 21
Velocity Estimates ................................................................................................................... 21
Velocity-Driven Development ......................................................................................................... 22
Some Considerations in Using Velocity Estimates ..................................................................... 23
Time Boxing .............................................................................................................................. 23
Gantt Charts.............................................................................................................................. 24
Release Plans ........................................................................................................................... 26
Module 3: Planning ......................................................................................................... 28
Estimating Task Time .............................................................................................................. 28
Cone of Uncertainty Principle ......................................................................................................... 28
Creating Time Estimates ................................................................................................................. 31
Task Dependencies.................................................................................................................. 35
Start-Start Dependency ................................................................................................................... 36
Start-Finish Dependency................................................................................................................. 36
Finish-Start Dependency................................................................................................................. 37
Finish-Finish Dependency .............................................................................................................. 38
Critical Path Method (CPM) Charts ........................................................................................ 38
Creating a CPM Chart ..................................................................................................................... 38
Critical Paths ..................................................................................................................................... 42
Program Evaluation and Review Technique (PERT) Chart ................................................. 43
Example of a PERT Chart............................................................................................................... 43
Critical Paths ..................................................................................................................................... 46

Agile Planning for Software Products | 2

 CPM Chart or PERT Chart? ........................................................................................................... 47
Iteration Plans........................................................................................................................... 48
Creating an Iteration Plan ............................................................................................................... 48
Sample Iteration Plan ...................................................................................................................... 50
Module 4: Risks ............................................................................................................... 53
Anti-Patterns ............................................................................................................................. 53
Group Anti-Patterns ......................................................................................................................... 54
Individual Anti-Patterns ................................................................................................................... 59
Causes of Failures ................................................................................................................... 61
Risk Assessment, Likelihood, and Impact ............................................................................ 62
Impact vs. Likelihood Matrix ........................................................................................................... 62
Risk-Value Matrix ............................................................................................................................. 63
Risk Strategies, Contingency, Mitigation .............................................................................. 64

Agile Planning for Software Products | 3

Course Overview

Upon completion of this course, you should be able to:

Understand the importance of planning.

Explain the role uncertainty plays in affecting plans and task estimates.
Break down tasks into appropriate sizes, and create a Work Breakdown Structure (WBS) for a
project.
Explain the differences between estimates, targets, and commitments.
Use different estimation techniques, including story points, velocity estimates and planning,
time boxing, and Gantt charts.
Generate work estimates for software products using a variety of approaches, including bottom-
up, analogy, experts, and the estimate formula.
Understand task dependencies, the different types of task dependencies, and how these affect
project planning.
Create Critical Path Method (CPM) charts and Program Evaluation Review Technique (PERT)
charts.
Create effective plans for software development on both iteration and release levels.
Identify anti-patterns and types of risks that may affect projects.
Assess and plan for project risks.

Introduction
Welcome to the Agile Planning for Software Products course, within the Software Product
Management Specialization. This course will focus on agile planning techniques. These
techniques ensure that the software product project is managed and organized properly while
also being adaptive to change.

The first two courses of this specialization, Software Processes and Agile
Practices and Client Needs and Software Requirements, serve as
foundational pillars in learning Agilethey are the legs of the inukshuk that
depicts the structure of our specialization. This course, Agile Planning for
Software Products serves as the body of the inukshuk, and rests on the
legs. This means Agile Planning for Software Products brings together
many of the fundamentals of those two prior courses and builds on them.

In the introductory course, you learned that building better software involves three goals. These
goals are:

Have the right product

Ensure it is done right
Ensure it is managed right

When a project is managed right, it adopts suitable practices to organize the work of everyone
involved and leads to clear next steps. This course focuses on ensuring a project is managed
right.

Agile Planning for Software Products | 4

There are three key types of planning used in software projects. This course will cover all three
of them. The types of planning are:

Release planning
Iteration planning
Risk planning

There are four modules in this course.

Module 1 will cover:

Major uncertainties that projects need to resolve
How to break down work into manageable pieces
The difference between an estimate, a target, and a commitment

Module 2 will cover how to:

Estimate the size of a requirement using story points
Measure a development teams productivity using velocity
Plan work using time boxing
Use a Gantt chart
Construct a release plan by mapping out the delivery of user stories over upcoming
sprints

Module 3 will cover how to:

Estimate the time it will take to finish a task
Determine task dependencies
Use Critical Path Method (CPM) diagrams and Program Evaluation Review Technique
(PERT) charts to plan work at the task level
Construct an iteration plan by mapping out tasks to be done within a sprint

Module 4 will cover:

Problems or risks that could cause a project to fail
Anti-patterns (situations with negative consequences for a project and its people)
Risk assessment and the level of action to take to address those risks
Constructing a risk plan by mapping out what to do should a risk occur

Agile Planning for Software Products | 5

Module 1: Introduction to Planning

Upon completion of this module, you should be able to:

Explain the importance of planning in a project.

Describe the key terms within Software Project Management: task, role, schedule,
milestone, work product.
Recognize and explain an Uncertainty Space (AUSP) diagram, and the concepts it
illustrates: means and ends uncertainty, and how they can evolve over a project.
Define work breakdown structure.
Recognize how the steps of project initiation, hardware/equipment acquisition,
development, testing, and implementation can be represented in a work breakdown
structure.
Create a work breakdown structure based on an example.
Identify the appropriate size for a task.
Explain the terms: estimate, target, and commitment.
Provide examples of estimates, targets, and commitments.
Understand what a good estimate entails.

Introduction to Planning
This course covers strategies and techniques for planning software projects within the Agile
manifesto. This means the strategies and techniques covered must leave the project adaptable
to change!

The key to planning software projects is breaking the project workload into small, manageable
tasks. This involves describing and organizing tasks, and assigning them time and work
estimates. Based on the identified tasks and their time and work estimates, an accurate
schedule for the project can be made.

Software projects can be scheduled many ways on release and iteration levels. Risk planning
must also be incorporated into creating schedules on both levels.

The release level refers to the stage where products delivered are ready for the market.
Release planning is used to plan the project as a whole. Common release planning techniques
include Gantt charts and release plans.

Iterations refer to smaller pieces of the project. They usually include a number of requirements
that must be finished. A release is made of many iterations. Iteration planning is a technique
that involves designing and developing the tasks that will be completed in a certain iteration of
the project. Common iteration planning techniques include PERT charts, CPM charts, and
iteration plans.

Both release planning and iteration planning must address potential risks that can arise in
software development. Risk planning involves preparing for those risks in order to avoid
negative repercussions in both creation and support of the product.

Agile Planning for Software Products | 6

Important Terms
A number of important terms are used over this course that are related to planning. These
should be well understood, before moving on to more complex discussions of planning software
projects within Agile.

Term Definition

Task A task is a small, manageable step of a project that must be completed. Tasks
are essential to a projecteverything in a project can be broken down into tasks.

Role A role is a duty that a person takes on or plays in some relation to the product, as
explained in the course Software Processes and Agile Planning. Examples of
roles include programmer, tester, designer, client, and product manager.

Work A work product is an output produced by a task or process, as explained in the

product course Software Processes and Agile Planning.

Schedule A schedule is created when the tasks of a project are mapped to a timeline.

Milestones Milestones refer to internal checkpoints used to measure progress. They are not
time based, but event or action based.
Milestones are most commonly used in linear and early iterative processes of a
project, as explained in the course Software Processes and Agile Practices.
In Agile, milestones are not generally used. Progress is instead measured by
working software as opposed to events or actions. Releases and iterations tend to
be time based as well, which do not fit with milestones.

These terms have certain dependencies on each other.

Roles perform tasks. For example, programmers write code for selected features.
Tasks not only create work products, but also might use them. This means that work
products might be dependent on one another. An output work product of a task might be
used as an input work product for a subsequent task. For example, the task of writing
tests produces the output work product of tests. These tests are then the input work
product for the task of executing tests.
Tasks, once identified, can be used to create schedules. Work and time estimates can
be assigned to tasks, and project schedules can be built from these estimates.

Understanding dependencies is important. Dependencies influence how tasks need to be

prioritized, and the consequential project schedules. For example, work products created in one
task that are then used as an input for a different task must be created before they are needed
in the project. Recognizing dependencies is therefore important for planning what to complete in
each iteration of the project.

Previous Concepts
A brief review of previous concepts helps show how previous courses relate to planning.

Agile Planning for Software Products | 7

Tasks, as described above, are an important part of planning a project, since everything in a
project can be broken down into tasks. Tasks also played important roles in previous courses.

In the course Client Needs and Software Requirements, you learned about creating
requirements. After the requirements have been created, they can be prioritized and broken
down into developer tasks to be completed. These developer tasks can be more technology
specific than requirements, which should generally be independent of technology. Developer
tasks are commonly scheduled using an iteration plan, as you will learn in this course.

In the course Software Processes and Agile Practices, you learned about creating
processes. Tasks are essential to creating processes, as they make up activities, and activities
make up phases. Each phase therefore has specific tasks that must be completed.

Tasks are also important to Agile practices. Agile practices provide ways of organizing tasks
and ensuring they are effectively completed. Iteration plans, as described above, are an Agile
practice and used commonly in Scrum methodology. Scrum is a popular methodology for
planning and organizing work on software products that follows the Agile philosophy.

Uncertainty Space
When beginning a project, there are many uncertainties. These uncertainties are often grouped
in two types. A well-managed project must identify and address both types, simultaneously. The
two types of uncertainty can be best understood through the questions:

What are you creating in the DID YOU KNOW?

product?
How will the product be developed? The grouping of project uncertainties into the two
types: means uncertainty and ends uncertainty
The first question is associated with ends was first proposed by the project management
uncertainty. Ends uncertainty refers to researcher, Alexander Laufer in 1997 in his book
uncertainty about what a product will do. Simultaneous Management. Laufers work is
The second question is associated with noted for using real stories from successful
organizations, so his project management
means uncertainty. Means uncertainty
theories are practice based.
refers to uncertainty in how a product will be
designed and developed.

Mike Cohn, one of the contributors to Scrum methodology, depicted the levels of uncertainty of
a product as it proceeds from concept, through development, towards an operational state in an
uncertainty space diagram. In such a diagram:

The means uncertainty should go from high to low, along the x-axis
The ends uncertainty should go from high to low, along the y-axis

Agile Planning for Software Products | 8

This is demonstrated the diagram below.

At the beginning of a project, the uncertainty for both what is being created and how it will be
created are very high, especially because requirements and methods will change over time. By
the end of a project, however, what will be created and how it will be created should be clear,
and uncertainties are low. At any point, if there is a high uncertainty for either axis, then some
uncertainty must still be solved for the product.

Navigating Uncertainty Space Diagrams

There are many different ways to navigate through projects using the uncertainty space
diagram:

1. Determine the what of a project first, then the how.

2. Determine how to develop a project, and then create it as you go.
3. Determine both the what and how of a project with as much balance between the two
as possible.

The first way identifies what the project will create and then how to create it. Using this method,
requirements are all determined first, and in great detail, before any plans for completing them
are developed.

Agile Planning for Software Products | 9

This method is likely following a waterfall process.

As illustrated above, this method leads to having low levels of ends uncertainty before the level
of means uncertainty changes. Often, this change is significant when it happens.

This method of uncertainty planning is not very adaptable because requirements are
established early and cannot change easily. It is therefore not in line with Agile methods.

The second way of navigating a project determines how to

develop a product and then creates the product as the Ad Hoc Development often
project moves along. Basically, the development team refers to development that is
determines implementation solutions and makes the product made up as you go.
up as they go.

This is ad hoc development, which involves little planning.

The chart above illustrates that second way leads to low levels of means uncertainty before the
level of ends uncertainty changes. As in the previous method, the change in uncertainty is
significant when it happens.

Agile Planning for Software Products | 10

This method of uncertainty planning can lead to wasted work if what is developed is not in line
with client needs.

The third way of navigating a project is somewhere between the first two. Both the how and
the what are accounted for around the same time, leading to balance between the two
extremes. This is a more efficient means of navigating uncertainty than either of the previous
two methods.

As demonstrated in the preceding graph, the levels of both ends uncertainty and means
uncertainty lower at an equivalent rate.

However, development is not always this straightforward, as sometimes requirements are

changed or added that will affect and create new uncertainties in both what the product will be
and how it can be built.

A Final Note on Uncertainty Spaces

In most real-world cases, clients and the development team will define the majority of
requirements for the project before planning begins. During this process, there might be less
ends uncertainty than means uncertainty. However, once product development begins and the
how is planned out more, the means uncertainty will lower.

As requirements are added or changed over the course of the project, there might be a slight
fluctuation towards lower ends uncertainty, followed by lower means uncertainty. Development
and solutions change as requirements change, so uncertainty will look more like the diagram
below, instead of a straight line.

Agile Planning for Software Products | 11

Both ends and means uncertainty should ultimately work their way towards being low. As
changes or added requirements happen and any corresponding changes in development follow
quickly, the project will stay adaptable and avoid wasted time and resources.

No matter the method of navigating uncertainty space, tasks are needed for the project to reach
its end.

Work Breakdown Structure (WBS)

Creating a software product involves many tasks, including determining requirements and
features, designing them, writing code, and testing. Often, other kinds of work are also involved,
such as marketing the product and arranging technical support and operations. All of these
require planningthe development team cannot simply begin developing the project without
any plans. The key to being able to plan a software project is to break it down into manageable
pieces.

In project management, the technique of work

breakdown structure (WBS) creates a visual Instructors Note:
representation of the process of breaking down large
A work breakdown structure is
project work products into smaller task work products.
similar to concepts you learned in
Tasks should be broken down until each is small and the course Software Processes
clear enough to be assigned to a single person. WBS is and Agile Practices. For
therefore also a hierarchical representation of project example, a process is broken
tasks and deliverables. It can take on the form of a list or down into activities, which in turn
other visual representation. Dates and order of execution are broken down into tasks.
are not included in a WBS.

Creating a work breakdown structure is a good first step in putting together a detailed plan for
the project. It is important to be balanced in breaking down tasks within a work breakdown
structure. Although tasks should be broken down until they are manageable, if they are too
small, this will lead to micromanagement.

Agile Planning for Software Products | 12

 TIP:

The acronym SMART is sometimes applied to determine if a task or work product is a good size. The
letters in SMART can stand for many things, but it generally means that a task should be:
Specific (the task should be clear and should not overlap with other tasks)
Measurable (the task should have a definite way of being marked as finished or done)
Achievable (one person should be able to do the task)
Relevant (the task should be within scope of the project)
Time boxed (the task can fit within a certain time frame)

Some guides to creating WBS suggest other more specific rules of thumb to help know if tasks are
manageable, including that the lowest level should take no more than 80 work hours to finish.

Another important tip is to remember to keep tasks within the scope of the project! This ties into the
100% rule often associated with WBS, which states that 100% of the work defined by the project
scope should be included in the WBS.

WBS can be organized functionally or by phase. Common steps included are:

Project initiation (when a new project is started and objectives, scope, purpose, and
products are decided)
Hardware and equipment acquisition (determining what hardware or equipment is
needed to create the product, and obtaining those)
Development (designing and creating the product)
Testing (determining the performance and reliability of the product to identify and fix
problems)
Implementation (when the product is executed for outsiders of the project to be able to
access and use)

These steps should be found at the highest level of the WBS (underneath the top level, which
should be the final product or project), and then broken down appropriately in subsequent
levels. A good resource for learning more about WBS levels is the Project Management Body of
Knowledge Guide (PMBOK) (see Course Resources).

In general, final tasks identified in the work breakdown structure should be independent and
unique of other tasks.

Example Work Breakdown Structure

Below is a graphic that illustrates how a work breakdown structure might look. In this example,
the project is to create a News Application.

Agile Planning for Software Products | 13

The product is the highest level of the WBS, and it is subsequently broken down into smaller
tasks in a hierarchical tree. The level under the highest shows steps similar to phases, including
programming and documentation.

Looking more closely at the Programming branch and its sub-branches, Links/Buttons,
Layout, and Live Embedded Widgets, you can see that each of these items can be broken up
into smaller tasks.

Agile Planning for Software Products | 14

For example, all three need a source code. Any work products will also have associated tasks in
the form of creating unit tests and acceptance tests.

Unit tests test the low-level functionality of source code; they should be written before source
code is. For the sub-branch Links, a unit test could determine if links are working
appropriately, for example. Acceptance tests, on the other hand, verify the application as a
whole. In this same sub-branch, an acceptance test could test the interface to make sure that all
the links are functional and correct, for example.

Designing the layout is another work product that should be listed, as well as any acceptance
tests associated with the design that make sure the layout is compatible and functional on
multiple devices.

These smaller tasks of creating designs and writing tests are good-sized work products. They
are much more manageable than a large task such as create a news application would be.
They are also in scope, unlike a task such as create entire source code for reader chat room.
A task like the last one should be not be included in a work breakdown structure, because it is
too large to tackle easily, and can easily result in missed smaller tasks and nuances.

Creating a Work Breakdown Structure

Large projects may have hundreds or thousands of tasks, and using a simple WBS tree chart
similar to our example can quickly become unwieldy. Many other tools are available to help
create work breakdown structures on larger scales, including Microsoft Excel and Microsoft
Project. Both are better suited for large-scale projects.

Work breakdown structures are also not limited to product management concerns surrounding
the Design and Implementation phase. They might also include Specification phase activities,
Verification and Validation phase activities, promotion, arranging technical support, gathering
analytics, setting up servers, etc.

Uses of Work Breakdown Structures

Work breakdown structures can be used many ways.

The most common use of a WBS is to determine tasks for the project based on the work
products displayed. Small, manageable work products can easily be turned into small,
manageable tasks. If the work product is Source Code for Links, as in the example above,
then the task Write source code for Links can be easily created. This is very helpful in the early
stages of planning a project because it considers the entire project, and tasks can then be used
to create schedules.

In a similar vein, work breakdown structures can be used to help assess how long it will take a
development team to complete a project. It is easier to estimate how long small tasks will take to
finish, and then to add those times together to determine a project estimate, than to estimate the
project as a whole.

WBS may also be used to determine potential risks the project will encounter. Once work
products have been identified through a WBS, risks can be determined for each of those work

Agile Planning for Software Products | 15

products. This is much more efficient than trying to determine risks from looking at the product
as a whole.

Finally, a work breakdown structure can also be used to demonstrate a product to a client in a
way that allows the client to look at features individually.

Estimates, Targets, and Commitments

The terms estimate, target, and commitment are often used interchangeably in project
management. But, in fact, all three of these terms mean something different.

Estimates
An estimate is a guess of the time it will take the development team or worker to complete a
task. A good estimate is usually based on a well-formed task, which adheres to SMART. They
are also usually determined by the developers who will work on the task.

An estimate should be as accurate as possible, and it is often based on data, like the amount of
time it took to accomplish a similar task. However, because a precise number of hours or days
is probably inaccurate, estimates are usually ranges.

It is important to understand that although estimates are sometimes presented as ranges, they
are not negotiable. This means that an estimate should only take into account previous work. It
should not include padded or extra time or be affected by what stakeholders or managers want
the estimate to be.

Other issues that might affect the ability to make a good estimation include:

Lack of developer experience with similar tasks

Too much attention to detail on tasks leading to over-estimates
Not enough information available on the task to create an accurate estimate

This last issue, lack of information leading to difficulties in creating a time estimate relates to the
Cone of Uncertainty Principle. This principle suggests that time estimates for features are less
accurate the further into the future those features are intended to be developed. For example,
estimating development is more difficult at the beginning of the project, when ideas are still
being passed around, than towards the end of the project, closer to actual development stages.

The Cone of Uncertainty Principle is explored in larger detail in the lesson Estimating Task
Time in Module 3: Planning of this course. It was also touched upon in the previous course,
Client Needs and Software Requirements.
Developers may also express reluctance in determining an estimate if they fear not being able
to meet an estimate. This could be a sign of larger organizational difficulties within the project.
Clients may assume that estimates are targets or commitments, which could be problematic.
Clear discussions about estimates, targets, and commitments between the development team
and clients helps address this issue.

Agile Planning for Software Products | 16

Approaches for creating an estimate include:

Bottom-up
Analogy
Experts
Using an estimate formula

More information on how to use these approaches can be found in the lesson Estimating Task
Time of Module 3: Planning in this course.

Targets
A target is a specific point in the schedule for the project or part of the project to meet. It is
almost an ideal deadline, and is usually set externally to the development team. Like estimates,
targets are not negotiated. Once a target has been decided, it must be adhered to for product
success and client satisfaction. In fact, targets are often contractual and can be used in
arranging product promotions.

Examples of targets are ends of a sprint or the release date of a product.

Commitments
Commitments are what you agree to deliver. Commitments are where negotiations can
happen. Based on both estimates and targets, a commitment can be negotiated. A good
question to ask when determining commitments is, How do I compromise on what the
development team will commit to doing, based on their estimates and the time constraints (of
the project)?

In order to determine a commitment, best practice suggests that estimates should be discussed
with the development team first, so accurate numbers can be determined. There should also be
discussions with the client in order to determine target dates. Then, based on these discussions,
commitments can be determined with both the development team and the client.

Estimate, Target, and Commitment Example

To better understand estimates, targets, and commitments, examining an example is beneficial.
A theoretical project may be estimated by the development team to need 1,000 hours of work,
based on the requirements the client has expressed. This is an estimate. The client, however,
may want the project finished within 600 hours. This is a target. After negotiating, the client and
development team may adjust the scope of the project so it seems to include about 500 hours of
work. This is a commitment. The project may take more or less time to finish than the
commitment, and that actual time can be used as data for future estimates.

As you can see in this example, although both estimates and targets are non-negotiable once
set, it is still possible to negotiate what you commit to for the scope of the project.

It is common for estimates to be automatically converted into commitments and targets, but this
is a bad practice. Drawing on the previous example, if a project is estimated to take 1,000 hours

Agile Planning for Software Products | 17

of work, often clients will automatically assume that this is a commitment of 1,000 hours. This
often leads to the temptation of inflating estimates, also a bad practice.

It is therefore important that estimates, targets, and commitments are made clear and separate
for both the client and the development team in their discussions. If this can be done, it is more
likely that the project and its target dates can run as planned, without the scope becoming too
large, as it ensures that target dates can actually be met. Again, drawing from the above
example, understanding that 1,000 hours is an estimate, and the target is 600 hours helped
create a commitment of 500 hours. This commitment helped narrow the scope of the project to
stay within the target.

Agile Planning for Software Products | 18

Module 2: Estimation

Upon completion of this module, you should be able to:

Define and create a story point.

Identify different ways of sizing a task.
Explain what velocity is, how to calculate a velocity estimate, what factors influence velocity,
and what makes velocity stable.
Explain the concept of done, and how a story is only counted towards velocity if it is done.
Define time boxing and its relation to Scrum.
Define release.
Explain and create a Gantt Chart.
Explain and create a release plan.

Story Points
As explored in the previous module, good estimates are based on good reasoning. For
example, the amount of time it took to develop and finish a task in the past could be the basis
for an estimate on the amount of time it would take to complete a similar task in the future.

In an ideal world, past performance would be the only thing that influenced an estimate. In
reality, however, developers often feel when they give an estimate that they should pad their
workin other words, some extra time should be addedin case their initial estimate is
mistaken for a commitment and is wrong. In fact, it is very difficult to make a reliable estimate,
especially if the task is far in the future (see the Cone of Uncertainty theory discussed in
Module 3 of this course).

Estimates are easily mistaken for commitments because they are both measured in units of
time. It is easy to jump to the conclusion that the time suggested in an estimate can be simply
converted into work hours. However, this isnt very accurate, and should be avoided.

Story points were developed as an alternate system for estimates. Instead of estimating how
much time work will take to finish, story points are an assigned value that look at how long it
will take to finish a piece of work in relation to other pieces of work. Story points are therefore
relative measurements, and unitless, unlike time estimates. In time estimates, hours or days are
the units used, whereas in story points, the points are not representative of any particular
measurement. This is explained more in How to Use Story Points, below.

How To Use Story Points

Story points build on concepts explored in the course Agile Requirements for Software
Products, such as user stories. User stories are short, simple, uniformly formatted descriptions
of a feature told from the perspective of the person who desires the feature. User stories
generally take on the format:

Agile Planning for Software Products | 19

As a _____, I want to _____, so that _____.

Another concept from the same course that is relevant to story points is the Scrum product
backlog, which is a set or list of software features that the software product manager and
development team plan to develop for a product over the course of a project.

In order to use story points, a relatively easy to estimate product requirement or user story
should first be selected from the product backlog and assigned an integer value. This value is a
story point. Using this basis, story points are then assigned to the rest of the user stories. For
example, if a task twice as big as the base task is found, it will be assigned approximately twice
the number of story points.

After this is done, you will be left with a list of user stories whose work estimates are all relative
to each other, as they are all based on the same point system. This prevents accidentally
committing to a time period, but instead just determines how large each user story actually is.
This is much more in tune with the true purpose of estimates.

The total number of story points can be added up for all of the user stories in the project, giving
some idea of how large the project will be.

It is extremely important, however, to remember not to treat story points like hours, or they offer
no advantages. For example, it is easy for a developer to just translate hours into user points.
In general, story points should not be very precise; instead they should stay relative. This
means that large numbers such as 10, 50, or 100 are bad as base numbers because they are
prone to be viewed as percentage points.

A good practice is to use story point values that are fixed. This helps avoid story points from
simply becoming interpreted as hours.

TIP

A good way to make sure story points stay fixed and relative is to use the Fibonacci sequence. The
Fibonacci sequence is a sequence where any individual number is the sum of the two numbers that
precede it. The sequence runs like: 1, 1, 2, 3, 5, 8, 13, 21, etc.

If the Fibonacci sequence is used, story points should be restricted to only those numbers for
estimates. This means that if an estimate falls between two numbers in the sequence, you must round
up the nearest available number. For example, if you have a user story that is approximately twice as
much work as one with the assigned value of 2, you should assign it the number 5, not 4.

Using the Fibonacci sequence for story points keeps estimates tidy and avoids making large estimates
unnecessarily precise. It also accounts for uncertainty this way. As it is not a direct correlation with
hours, using the Fibonacci sequence also prevents unintended commitments from being formed.

Agile Planning for Software Products | 20

Advantages of Story Points
Aside from the face that using story points as estimates helps to not create unintended
commitments, story points offer other advantages. Creating estimates using story points tend to
lead to more accurate estimates. Humans in general tend to be bad at making exact estimates
from scratch, but a method of estimating based on relative sizes makes it easier.

Imagine trying to estimate the size of a building. Your gut feeling might tell you that the building
is sixty metres tall. Unless you actually measured the building, you would have no idea of how
tall the building exactly is. So, your estimate could be very inaccurate. If you were to use a
system similar to story points, however, you would compare that building to another, smaller
building nearby. If you assigned the small building a value of 1, and you believe the taller
building to be about three times taller, then it would have a value of 3. It is must easier to
estimate this way than to just use a best guess.

Using story points for estimations is also potentially less stressful for developers, because
nothing is exact. This means it is harder to go over the estimate like one could go over a time
deadline.

Limitations of Story Points

Although story points offer many advantages, they are not a widely accepted practice, even for
those who use Agile methods and Scrum. This is because there are several notable limitations
to story points too.

One way in which story points can be difficult is that they require a shift in thought from time to
points. As a result, developers might be tempted to treat points like hours. This limitation could
be resolved through increased exposure and practice with using story points.

Estimates are always difficult to make accurately, whether with

hours or story points. Although using relative sizes generally As an example of how bad humans
results in more accurate estimates than using exact numbers, are at accurate estimation, consider
this is still open to mistakes. the vertical-horizontal illusion. It can
be difficult to tell if a horizontal line
and a vertical line are the same
length, even if both are visible at the
same time.

Story points are open to limitations on large teams as well. Sometimes, story points are inflated
to give the appearance of productivity. This practice is acceptable if its applied consistently
(e.g., if the inflation of the numbers is the same across the project). If inflation of numbers starts
halfway through the project due to the introduction of new features or pressures, then story
points will lose value and become less useful for estimation. The important thing about story
points is consistency.

Velocity Estimates
Velocity is a measurement of the work that can be taken on and finished over the length of a
sprint. Velocity is represented as the number of story points completed over a given period of

Agile Planning for Software Products | 21

time. An example of a velocity could be 15 story points per sprint. Velocity is helpful because it
serves as a means of:

Estimating the effort the development team can achieve throughout the project
Estimating effort that can be achieved on future work
Indirectly tracking the teams productivity

Velocity can be measured as the number of user stories

In this course, a sprint is the same or work items delivered over a sprint. However, using user
thing as an iteration they are both stories in velocity can be problematic, because not all
short, iterative, and incremental time user stories in a project are the same size. Velocity can
periods. In other words, both are
also be measured through the number of story points
ways to time-box work in small
chunks. Time boxing is explored completed over a sprint. Story points make for more
more in the next lesson. accurate measurements.

A project should use either user stories or story points in

velocity and consistently use the same measurement. Mixing the two would be confusing, as
they are interpreted in different ways. In this course, we will focus on using story points over
sprints as our productivity metric.

For story points to count towards velocity, they must be finished completely. Partially finished
tasks do not count towards velocity. Using story points that are considered completely done in
velocity estimates is very important.

How can it be determined if a user story is done? In Scrum and Agile methods, a user story is
not done when the functionality has only been coded. A user story must also pass both unit
tests and acceptance tests and then must be documented before it is considered done. The
additional completion of testing and documentation to coding is sometimes referred to as done
done in the field.

Velocity-Driven Development
Velocity serves as the foundation of velocity-driven development. This kind of project
development happens when the velocity of previous sprints is used as a basis for planning other
sprints in the project. For example, if the three previous sprints of a project had a velocity of 15
points per sprint, then this can be used to plan the next sprint. User stories can be selected
based on their story points until the total number of points projected to be finished in a sprint is
15 points or less. A development team knows better how much work they can get done if they
can use past data to help predict what might be done in the next sprint.

Velocity-driven development requires a certain amount of stability in past velocity. At the

beginning of a project, velocity may not yet be steady as there are still a certain amount of
inaccuracies, and it may take a few sprints for it to even out. In order to help calculate stable
velocities, it is important to be strict about counting story points only when they are done. For
example, if two stories worth 3 points each are complete at the end of the sprint, and a third
story worth 4 points still needs testing, this third story should not be counted. The sprint total
should be 6, and not 10. This practice also helps to force or to encourage closure to stories
rather than to constantly start new ones.

When estimating velocity, will find yourself in one of two places:

Agile Planning for Software Products | 22

 The beginning of a project, when it is difficult to estimate velocity because no prior work
has happened on which estimates can be based, or
The middle of a project, when it is easier to make estimates based on prior work.

As in estimation, in cases where no prior work can be drawn on at the beginning of a project, it
is possible to:

Consult experts who may have worked on similar projects previously and who might be
able to offer insights, or
Draw on the teams own prior experience with similar projects to make estimates.

Previous data can be used to determine estimates through a variety of techniques, including but
not limited to:

Linear extrapolation (extending graphed data beyond the known data in a straight line
based on how previous data has behaved),
Averaging the velocity of the three previous sprints, and
Using the minimum velocity of the three previous sprints.

Some Considerations in Using Velocity Estimates

There is a little bit of controversy over using velocity estimates that are important to
acknowledge. Velocity estimates are not necessarily accurate, because velocity is variable. This
reinforces the importance of only counting stories that are done by the end of the sprint. This
helps counteract variability. It is also important to remember that velocity is derived from
estimates, so it should still be seen as an estimate.

Velocity still offers many advantages, as it shares the advantages offered by story points.
Velocity is also considered better for long-term tracking, especially as velocities become more
consistent over the course of a project. And although velocity is not considered an ideal means
of tracking the individual productivity of team member, it is considered a good tool for decision-
making.

For more information on some of the considerations on using velocity estimates, see Michael
Cohns Why I Prefer Commitment-Driven Planning in the Course Resources.

Time Boxing
Time boxing is creating a restricted time period in which something is to be built, or for certain
tasks to be finished in. The key to time boxing is the strict time limitthe event must stay within
a predetermined timeframe.

Time boxing is a popular organizational method, as it allows software teams to

compartmentalize work they have planned. Good time boxing also plans out the work at the
beginning of a time period and leaves room for reflection on the progress of the project. Time
boxing is particularly important to Scrum.

Agile Planning for Software Products | 23

Scrum sprints, usually between two to four weeks long, a type of time box. They consist of a
relatively small amount of work that is restricted to a relatively short development time. Scrum
sprints include planning and review periods before and after work is done.

Time boxes are created when a time period is set in the project during which a certain amount
of work will be planned, accomplished, and reviewed. Sprints and iterations are a kind of time
box. Time boxing helps keep a project on track at a regular, sustainable pace because it divides
the project into manageable chunks. The goals and milestones generated by time boxing helps
development stay within scope and keep things on schedule. As work is planned over small
doses of time, determining upcoming targets and their time periods becomes easier. It makes a
project much more efficient than working with an uncertain schedule.

Developers should beware of putting too many tasks into a time box, or they cannot be finished.
Velocity helps inform how much work can be completed in one time period, as previous work
informs new estimates.

The organizational ability time boxing offers projects is sustainable. Further, it allows the team to
learn from work at regular intervals, because of the built-in review period. It also allows the
development team to see how work will be divided throughout the project.

A project can be released in stages, at the end of each time box. This is because at the end of
each time box, working software should be created, which can be released as a product. All
work in a time box must be finished. Time boxing thus ensures that a tangible product is
available at the end of a project, even if resources are lost along the way. This kind of progress
also improves team morale.

Gantt Charts
Gantt charts are simple tools for visualizing project plans or schedules. They are a kind of
cascading horizontal bar chart. Gantt charts can also be used to visualize task dependencies,
the length of tasks, and deadlines.

Gantt charts plot tasks on the left-hand side of the chart against dates, which are displayed at
the top of the chart. Each task is listed separately, one by one, and given its own horizontal bar.
The first scheduled task is at the top of the chart, and progressive tasks are displayed
underneath it. Each tasks bar represents the duration of that task from its start date to its finish
date. Once all tasks are laid out on a Gantt chart, you can see when each individual task is
scheduled to begin and end.

Below, is a rough template of what a Gantt chart looks like. In a real example, the tasks would
be given descriptive names, and there may be several more tasks.

Agile Planning for Software Products | 24

 Example Gantt Chart
31-Dec 10-Jan 20-Jan 30-Jan 09-Feb 19-Feb 01-Mar 11-Mar

Task 1

Task 2

Task 3

Task 4

Task 5

Task 6

Gantt charts are useful in both Agile and non-Agile projects. Because sequential tasks can be
displayed in a Gantt chart, non-Agile processes, such as the Waterfall process, can make
productive use of Gantt charts. In Agile environments, Gantt charts are also useful.

In order to use Gantt charts successfully with Agile methods:

Additional context should be provided to tasks (such as identifying if there are task
dependencies)
Parts of the chart can be sectioned off as sprints
The charts must be used with an understanding of flexibility and adaptation to project
needs (in other words, the chart could change as the project unfolds)

As detailed on the preceding page, Gantt charts allow the team to visualize when a project will
be finished, when tasks are to be worked on, the length of those tasks, and potential project
dependencies. Project dependencies and the different types of dependencies that can exist in
projects are described more in the Task Dependencies lesson of the next module, but a brief
outline of representing task dependencies in Gantt charts follows below.

In order to display task dependencies on a Gantt chart, first, a project backlog of tasks should
be created and prioritized. Tasks that are dependent on the completion of other tasks should be
identified on this list.

When a Gantt chart is produced based off of tasks and their time estimates, tasks that are
dependent on others should be identified through the use of an arrow, going from a task
towards the one that depends on it. In general, however, other planning techniques such as
Critical Path Method (CPM) charts and Program Evaluation Review Technique (PERT) charts
are favoured for demonstrating task dependencies. More information on how to visually
represent task dependencies can be found in Module 3: Planning of this course.

Agile Planning for Software Products | 25

In the example below, Task 5 is dependent on Task 4, as represented by the flowing arrow.

Example Gantt Chart with Task

Dependency
31-Dec 10-Jan 20-Jan 30-Jan 09-Feb 19-Feb 01-Mar 11-Mar

Task 1
Task 2
Task 3
Task 4
Task 5
Task 6

Gantt charts can be used on both the iteration level or the release level. On the iteration level,
Gantt charts help provide insight by outlining tasks that should be completed within a few days
in a current sprint. On the release level, Gantt charts represent user stories instead of tasks, so
it can provide information on user stories over an upcoming series of sprints.

Many different tools exist to help create Gantt charts, including:

Microsoft Excel
Microsoft Powerpoint
Gantt Project, an open-source Gantt chart creator (http://www.ganttproject.biz/)
Many other subscription projects

Release Plans
Planning is a process that can occur on both the iteration level and the release level. This is
known as two-level planning. As explained in the first module of this course, iteration planning
involves dealing with tasks on a detailed level, while release planning deals with broader level
user stories. Both levels of planning rely on time boxing in the form of sprints to organize either
the tasks or user stories.

Iteration plans are generated at the beginning of a sprint in Scrum and Agile methods during
the scheduled planning event. The iteration plan plans out the developer tasks to be completed
during that particular sprint. Tasks tend to be actionable items, or items that a developer can
complete. A user story consists of many tasks. Tasks are then self-assigned by the developers
on the team before the work even begins. At the end of the sprint, working software should be
created, which can then be shown to the client during the review event. Iteration plans are
explored in more detail in the next module.

Agile Planning for Software Products | 26

Release plans, on the other hand, are used to plan a few
sprints at a time by determining which user stories from the TIP
project backlog should be completed and released in which
sprint. User stories are chosen by priority. Generally, release In addition to creating release
planning occurs before any sprint is startedtherefore, it plans that cover a few sprints,
occurs before iteration planning. User stories in release plans the product manager should
also create a long-term road
should be distributed logically and evenly across sprints; map plan, which covers the
otherwise, they may be finished in a disorganized or majority of the project and its
haphazard manner, which may cause issue with task user stories, if not the entire
dependencies. project. This helps give an idea
of how the project will
A prioritized project backlog, as described in the Client progress.
Needs and Software Requirements course is therefore key
to release plans. User stories can be planned in a release
plan from the top priority down:

1. Must do user stories (top priority)

2. Should do user stories (medium priority)
3. Could do user stories (low priority)

User stories are distributed across sprints pulling from these priorities, until each sprint is filled
with an appropriate amount of work.

The number of user stories to assign during a sprint can be determined through many ways,
including:

Prior experience can help inform how many user stories can be finished in a sprint
Data such as velocity estimates and story points can be used. Velocity can inform how
many story points might be able to be finished in a sprint. User stories can be added
until their story points add up to more than the projected limit.

Release planning should give an accurate projection of what the product will look like at the end
of an upcoming series of sprints. A good release plan helps with client expectations and
increases developer morale because tangible results are planned and expected.

Agile Planning for Software Products | 27

Module 3: Planning

Upon completion of this module, you should be able to:

Describe what makes a good estimate.

Explain issues that may affect estimation, including the concept of the Cone of Uncertainty.
Explain different approaches for creating estimates, such as bottom-up, analogy, and
experts.
Be able to calculate task estimation time using the estimation formula.
Understand task dependencies and the different types of task dependencies: start-start,
start-finish, finish-start, and finish-finish.
Explain a critical path method (CPM) chart.
Explain the concept of slack.
Explain and create a PERT chart.
Explain the differences between a CPM chart and a PERT chart.
Explain and create an iteration plan.
Understand that planning is self-assigned work done by developers; it is not assigned by
clients.

Estimating Task Time

Cone of Uncertainty Principle
Before exploring the different approaches that can be used to estimate task time, it is important
to understand the Cone of Uncertainty Principle, which informs the accuracy of an estimate.

The Cone of Uncertainty Principle works off of the fact that uncertainty is high at the beginning
of a project and becomes lower as the project progresses, as discussed in the lesson
Uncertainty Space in Module 1: Introduction to Planning of this course and in the course
Client Needs and Software Requirements. The principle suggests that estimates made when
uncertainty is high will be inaccurate. It also suggests conversely that as the project progresses,
uncertainty is reduced because the requirements and development pace become clearer and, in
turn, estimates will become more accurate.

The Cone of Uncertainty can be illustrated graphically.

Agile Planning for Software Products | 28

 Image adapted from Software Estimation: Demystifying the Black Art (2006)
by Steven C. McConnell. Used with Permission.

On this graph, the progression of the project over time is represented on the x-axis. Variability is
represented on the y-axis. Variability refers to the factors or the extent by which an estimate
could vary. The larger the factor, the greater the variability. Each number on the y-axis therefore
represents a factor of variability, and the x represents the mathematical operation of
multiplication.

The graph illustrates how at the beginning of a project, there is high variability and towards the
end of the project, the variability significantly lowers. This creates a cone shape, which is where
the Cone of Uncertainty principle gets its name.

The Cone of Uncertainty graph is illustrated below, with important points in the project
included. This helps illustrate how variation changes at different points in the project over time.

Agile Planning for Software Products | 29

The important project points are as follows:

The Initial Product Definition when the idea for the product is first formed. At this
stage, variability is high (between the approximate factor of 3 and 0.375), and estimates
will need a large range.
The Requirements Elicitation activity when the needs of the product and features
have been identified. At this stage, variation is smaller (between approximately 2 and
0.5), and estimate ranges will be smaller too.
Formulating Potential Approaches when the methods used to do the project are
chosen. At this stage, variation (between the approximate factor of 1.25 and 0.8) and
estimate ranges become even smaller.
The Architecture Design and Detailed Design stages, when information is known on
how and what will be created in the end product. At these stages, the estimate becomes
even more reliable as variation decreases (variation becomes closer and closer to a
factor of 1).

Estimate ranges for the project can be made using the variability as illustrated in this graph. In
order to do this, the estimate should be multiplied by its corresponding factor variability for
where you are in the project. These will make the top and bottom numbers of the estimate
range. For example, if a task is estimated to take 6 hours in the Requirements Elicitation Stage,
this means it has a variability of 2 and 0.5 on the y-axis.

2 x 6 = 12
0.5 x 6 = 3

This gives an estimate range of 3 to 12 hours. As you can see, at this stage, there is still a great
deal of uncertainty.

Using variability to create estimate ranges helps ensure a more accurate range. Estimates
should also be continually revised as the project moves through the Cone.

Agile Planning for Software Products | 30

Creating Time Estimates
There are many approaches to creating time estimates. All of these approaches, however,
should be based on some sort of data. Remember also that estimates are non-negotiable
figures.

This lesson will present several approaches that can be used to create time estimates:

Bottom-up TIP: If it is necessary to create an estimate quickly,

Analogy usually the bottom-up technique is the best to use,
Experts if enough information is available.
Using an 3-Point Estimate Formula

Bottom-Up
Bottom-up is a technique for creating a time estimate that involves breaking apart a project into
small, manageable tasks, as done in a Work Breakdown Structure (WBS). Once the tasks have
been broken up, they are much easier to estimate. If all the small tasks are given estimates,
these estimates can then be added together to create an estimate for the full project.

Below is an example of a Work Breakdown Structure (WBS), where each of the tasks has been
represented as hours. Once all the hours are added up, we can see how long each section of
the structure is estimated to take and how long the entire project is estimated to take.

Agile Planning for Software Products | 31

Note that even small tasks, which are easier to This technique assumes that none of
estimate, can be difficult to estimate accurately if there the tasks in the Work Breakdown
is not enough data to base that estimate on. This Structure (WBS) can be performed
technique therefore works well when combined with simultaneously, but instead that each
task must be completed before another
another, especially in cases where there is not enough
can start.
data to create an accurate estimate.

Analogy
Analogy is a technique for creating a time estimate that works by comparing the current task or
project with a similar experience. If a similar task or project was accomplished in the past with
similar scope and costs, then it can be a good basis for estimating the current project.

This technique should be used with some caution, as it can only work if the same team has
worked on a similar project. This means the work, process, technologies, and team members
have to be the same or similar to the current project to be a meaningful basis of data.

Experts
Time estimates can also be generated through experts. This technique works by converging
estimates made from multiple estimatorsgenerally developers who have prior experience with

Agile Planning for Software Products | 32

the kind of work at hand. This technique is normally used in cases where the development team
has limited experience with the work required for the project.

This technique can be costly, as it requires several estimators to look at the project. It can,
however, also be very accurate.

TIP: It is important to remember that all projects will have unique issues that arise over the course of
the work, regardless of how similar it is to other projects. This means that any of the bottom-up,
analogy, or experts techniques should be made with the understanding that these unique issues,
which could not have been otherwise accounted for, may affect estimates.

The PERT 3 Point Estimate Formula

A useful tool used to estimate task time more accurately is the three-point estimation. This is
known as the PERT 3 Point Estimate Formula, and it can create a range estimate. The
formula is based on three values, which can be determined through prior experience or best-
guesses:

Most Probable Time (Tm) is the most likely estimate of the time it would take for the
task to finish. This estimate operates under the assumption that the project or the task
may encounter normal problems. Any of the bottom-up, analogy, or experts techniques
can be used to generate this estimate.
Optimistic Time (To) is the best-case scenario estimate. It assumes that everything will
go right over the course of the project or the task and that the development team will
work efficiently. It is therefore the least amount of time the task or project could be
completed in.
Pessimistic Time (Tp) is the worst-case scenario estimate. It assumes everything or
almost everything that could go wrong over the course of the project or task will. It is
therefore the maximum amount of time the task or project could be completed in.

Using these three variables, the Expected Time (Te) can be calculated for the task. The
Expected Time is a weighted average that can be calculated with this formula:

Te = (To+4Tm+Tp)
6

This formula places emphasis on the Most Probable Time that is multiplied by 4, giving it more
weight in the formula. It is then averaged with the Optimistic Time and the Pessimistic Time.

The Expected Time is usually represented as a range. In order to create a range, we need
some information about the probability that the estimate is correct. This is done by calculating
the deviations of the Expected Time.

Agile Planning for Software Products | 33

Deviations are represented by the Greek letter sigma (). Standard deviations represent how
measurements can be spread out from the expected value. These are based on normal
distribution plots or bell curves, which work based on percentages.

A normal distribution assumes that numbers are spread out in a certain way from standard
deviation (in this case, the Expected Time), and that those values within one deviation () of the
standard deviation are 68.3% likely to be accurate. Within two deviations (2), there is a 95.5%
chance the number is accurate. These numbers are drawn from bell curve and statistic theory.

To calculate deviation, the following formula is used:

= (Tp-To)
6

The smaller the difference between the Optimistic Although deviations are used to create time
Time and the Pessimistic Time, the smaller the ranges in the course of planning, deviations
can also be expressed using the (plus or
deviation will be. Once the deviation has been
minus) symbol after the Estimated Time.
calculated, it can be subtracted from the Estimated For example, if the Estimated Time is 12
Time (Te) to get the low number of the range. The hours, and the deviation is 2, this could be
deviation can also be added to the Estimated Time expressed as 10-14 hours, or in some
(Te) to get the high number of the range. cases, 12 hours 2.

If one deviation () is used to create a range, we can assume that the range is 68.3% likely to
be accurate. However, if it is more important for the project to have an accurate estimate than a
smaller one, you can double the deviation (2) by multiplying it by 2, to reach 95.5% accuracy.
More information on the estimate formula can be found in supplementary readings found in the
Course Resources. A worksheet is also available to help with practice using the formula.

Sometimes a triangular distribution is also calculated using the same three variables, and the formula
is:

Te = (To+Tm+Tp)
3

This formula places less emphasis on the most likely scenario, as it not weighted higher by being
multiplied by 4. Using the variable in the example below, this equation gives us an estimate of 10 days
to finish the task, instead of 9. Agile Planning for Software Products | 34
Estimate Formula Example
Understanding the estimate formula can be easier through the use of an example. Using the
hypothetical numbers for a task:

To = 5 days
Tm = 8 days
Tp = 17 days

We can calculate the answers to the estimate equation:

Te = (5+ (4x8) +17)

6

Te = (5+32+17)
6

Te = (54)
6

Te = 9 days

In this example, standard deviation would be:

= (17-5)
6

= (12)
6

=2

So, the estimation for this task would be 9 days 2 (a plus or minus deviation of 2 days). This
would likely be expressed as an estimate of 7 to 11 days to finish the task.

Sometimes the numbers created through using the estimate formula will not be whole numbers (in
other words, they may have numbers after the decimal). In these cases, it is important to round the
estimate up to the next highest number. For example, if the formula yields the number 6.9, it should
be rounded up to 7. Similarly, if it yields the number 5.1, it should be rounded up to 6.

Task Dependencies
Task dependencies, as briefly discussed in the Software Processes and Agile Practices
course, and the lesson Introduction to Planning at the beginning of this course, refer to those
situations where one task is dependent on another task or set of tasks in some way.

Task dependencies are very important to planning and scheduling, because they shape the
order in which tasks must be done.

Agile Planning for Software Products | 35

There are four types of dependencies. They differ in how the second depends on the first:

Start-start dependency
Start-finish dependency
Finish-start dependency
Finish-finish dependency

Start-Start Dependency
Start-start dependencies refer to those dependencies where the first
task must start before the second task can start. When tasks finish in
this dependency is not important.
Start-start dependencies are visually represented as follows, with an
arrow going from the beginning of Task A to the beginning of Task B.
The two tasks are shown more or less in parallel. Task B depends on
Task A starting.

This method of representation can be used in any number of visual representations, including
Gantt charts and PERT charts.

An example of a start-start dependency could be working on a project and creating a glossary.

The project must be started before a glossary can start. These two tasks can be worked on
simultaneously, however, and it does not matter which one is finished first.

Start-Finish Dependency
Start-finish dependencies refer to dependencies where the first
task must begin before the second task can finish. These
dependencies generally involve some kind of overlapping handover.
In start-finish dependencies, the start of the second task is not
important. In other words, the first task can start before or after the
second task starts.

Start-finish dependencies are visually represented as follows, with

an arrow going from the beginning of Task A towards the end of
Task B. Both tasks are shown more or less in parallel. Task A must start before Task B finishes.

This method of representation can be used in any number of visual representations, including
Gantt charts, and PERT charts.

An example of a start-finish dependency could be planning the next sprint during the current
sprint being worked on. This allows the development team to begin working immediately upon
starting the next sprint. Planning (Task B) must start before the current sprint (Task A) ends, so
it is start-finish dependent on the current sprint.

Agile Planning for Software Products | 36

Finish-Start Dependency
Finish-start dependencies refer to
dependencies where the first task must
finish before the second task can start. This
is the most common type of dependency.

Finish-start dependencies are visually

represented as follows, with an arrow going
from the end of Task A to the beginning of
Task B. Task B does not overlap with the
Task A bar in any way. Task A must finish before Task B can begin.

This method of representation can be used in any number of visual representations, including
Gantt charts and PERT charts.

An example of a finish-start dependency could be the need to finish designing a feature and
knowing what it will do before a user or training manual can start to be written for a product.

Agile Planning for Software Products | 37

Finish-Finish Dependency
Finish-finish dependencies occur when the first task must
finish before the second task can finish.

Finish-finish dependencies are visually represented as

follows, with an arrow going from the end of Task A to the
end of Task B. The two tasks are parallel to each other. The
arrow shows how task A must finish before Task B can
finish.

This method of representation can be used in any number

of visual representations, including Gantt charts and PERT
charts.

An example of a finish-finish dependency could be finishing billing a customer for a product

before finishing handing over the product, in order to avoid not being paid for the work.

Critical Path Method (CPM) Charts

Task dependencies are very important and, as mentioned in the previous lesson, they affect
schedules and planning techniques. Critical Path Method (CPM) charts are an example of
such a task planning method. A CPM chart is a visual way to organize task dependencies.

Creating a CPM Chart

The steps involved in making a CPM chart are:

1. Make a list of tasks needed to finish the project and then create a Work Breakdown
Structure (WBS) for those tasks (see the Work Breakdown Structure lesson in Module
1: Introduction to Planning of this course).

2. Add time estimates to each task.

3. Organize the tasks by grouping dependencies horizontally together. For example, a task
that is finish-start dependent on another should be grouped next to the task it is
dependent on. Arrows can then be used depending on which task dependency is at work
to depict the dependency (see Task Dependencies lesson for how to graphically
represent different types of task dependencies).

Here is an example of steps 1 through 3 for a project whose goal is to create a spaghetti and
meatball dinner with a side salad.

Agile Planning for Software Products | 38

 4. From these task groupings, paths can be created. A path is a sequence followed using
arrows that you can take from one task to another (with dependencies shown as
appropriate).

Paths that go from the beginning to the end of the chart are called beginning-to-end paths.
Tasks on different beginning-to-end paths can be completed independently and, therefore, at
the same time.

If two or more beginning-to-end paths meet, this is a coordination point for task
dependenciesit means that the two or more tasks must be finished before the task they point
to can begin. In other words, the tasks must be coordinated.

A CPM chart can have as many beginning-to-end paths and coordination points as needed to
finish the project.

The example below shows a number of beginning-to-end paths generated from the Work
Breakdown Structure (WBS) from steps 1 through 3 of the spaghetti and meatball dinner
project. These paths illustrate the groupings under Pasta and Meatballs in the WBS.

Agile Planning for Software Products | 39

There are four paths in this example. Following the arrows, these four paths are from:

Boil the Water through to Add Meatballs to Dish

Heat the Pan through to Add Meatballs to Dish
Prepare Ingredients through to Add Meatballs to Dish
Pre-heat Oven through to Add Meatballs to Dish

There are also three coordination points:

When Strain the Pasta and Heat the Sauce coordinate before Put Sauce on Pasta
When Form the Meatballs and Pre-heat Oven coordinate before Bake the Meatballs
When Put Sauce on Pasta and Bake the Meatballs coordinate before Add Meatballs
to Dish

If an overall time estimate is desired, it can be calculated by adding together the longest
beginning-to-end path in a CPM chart. This is because all other beginning-to-end paths can
happen at the same time. In the example above, Pre-heat Oven can occur at the same time as
the path Prepare Ingredients Mix Ingredients Form the Meatballs.

A beginning-to-end path can also illustrate tasks that occur in parallel. Parallel tasks can occur
in any order, as long as they are finished before the final task.

Drawing on the Caesar Salad grouping from the WBS example above, it is possible to
illustrate parallel tasks.

Agile Planning for Software Products | 40

In this example, there are actually three beginning-to-end paths:

Add Croutons
Add Cheese
Add Dressing

These can occur in any order, as long as they are done before the salad is tossed. They are in
parallel.

5. Combine all of the beginning-to-end paths in order to create a full CPM chart.

Here is the full CPM chart for the spaghetti and meatball dinner project.

Agile Planning for Software Products | 41

Critical Paths
CPM charts, once completed, are a useful tool for generating a schedule for the project and
determining minimum time estimates. The major way this is done is through the use of critical
paths.

A critical path is the longest duration path of tasks between logical points. In other words, a
critical path is the beginning-to-end path with the greatest time estimate. It represents the
minimum amount of time needed to complete a project, assuming there are enough resources
that other beginning-to-end paths can be done at the same. The critical path in the example
CPM chart above has been highlighted in blue.

In addition to providing the longest time estimate for the project, critical paths also provide
information for longest and shortest time estimates for project tasks. Critical paths highlight the
tasks that are critical for the project. These are the tasks along the longest duration path, which
represent the minimum time for those tasks. But critical paths also highlight which tasks have
slackthe ones on a different beginning-to-end path from the critical path. Because other
beginning-to-end paths from the critical path have smaller time estimates, tasks on those paths

Time estimates generated from critical paths in CPM charts may differ from estimates generated from
Work Breakdown Structures (WBS). In a CPM chart, the critical path determines the minimum
estimate and it is assumed (where resources are available) that other paths will be worked on at the
same time. In a WBS estimate, however, each task is added as if one is done after another.

can be delayed, as long as the total estimate of the path never exceeds the critical path. This
potential extra time for tasks is what gives the estimate slack.

Agile Planning for Software Products | 42

Critical tasks in the critical path should be the first to be scheduled from a CPM chart, because
tasks in other paths can be given extra time to finish (within the critical path time limit), without
adding to the overall project.

Slack is useful to projects, as it provides a degree of flexibility to certain tasks, without

necessarily affecting the target completion date.

Program Evaluation and Review Technique (PERT) Chart

Another task planning tool affected by task dependencies is the DID YOU KNOW?
Program Evaluation and Review Technique (PERT) chart.
Similar to CPM charts, PERT charts are visual representations PERT charts were
of a project. developed in the 1950s by
the United States Navy.
PERT charts use nodes and edges in their network or Their original purpose was
graphical depictions. to manage the Polaris
submarine missile program!
Nodes, usually illustrated as circles, represent milestones.
Milestones are not time based, so in a PERT chart, nodes tend to represent some kind of event
such as the production of a work product or some other event occurring. In contrast, nodes in
CPM charts represent tasks.

Edges, illustrated as lines that connect nodes, represent tasks. In a PERT chart, after lines are
drawn, they are further denoted with tasks and time estimates. The lines are also given arrows,
to illustrate task dependencies (see the Task Dependencies lesson in this module) and the
order tasks must be completed in. A sequence of dependent tasks is therefore represented in a
path that follows the direction of the arrows.

If multiple independent paths come out of a node (that is, if multiple lines
come out of a node), it means that those tasks can be completed at the
same time, assuming time and resources are available. They are in
parallel.

Alternately, if multiple paths lead towards a node (if multiple arrows point to
a node), it means that these paths must synchronize. In other words, those
paths must be completed around the same time, before the next path
leading out of the node can start.

In general, PERT charts make dependencies easier to see.

Example of a PERT Chart

This example draws on the project discussed from the Critical Path Method Charts (CPM
Charts) lesson preceding this one, with the hypothetical project whose goal is to create a
spaghetti and meatball dinner with a side salad.

Agile Planning for Software Products | 43

The paths for making pasta would look as follows in a PERT chart:

Edges, or the arrows, have been labelled with task and time. Task dependency is clear through
the order of the arrows. Nodes, or milestones, have been labelled with numbers. Labelling
nodes this way is effective, as numbers are a simple system, but it also allows for nodes to be
easily referenced. Each node occurs after at least one event is completed.

Note in this section that there are two independent paths: 1 3 5 can be worked on during
the path 1 2 4 5.

The paths for meatball-related tasks would look as follows in a PERT chart:

The paths for salad-related tasks would look as follows in a PERT chart:

Agile Planning for Software Products | 44

Note the parallel tasks here, of Add Croutons, Add Cheese, Add Dressing. These tasks can
be done in any order, but all must be completed before the final task of Toss the Salad can
begin.

Once all the paths have been combined, the complete PERT chart for this example would look
like this:

As with building a CPM chart, in a PERT chart, each grouping of tasks should be charted first
and then those sections combined. In a PERT chart, this may require cleaning up numbers that
overlap for the same task in different sections. For example, node 6 above combined with node
11 in the final chart, and was simply labelled 11 (see below).

Agile Planning for Software Products | 45

 Nodes can be identified using any ordering scheme desired. However, using numbers increasing from
left to right allows one to refer to paths in an increasing order.

Critical Paths
Critical paths are very important in PERT charts. As in CPM charts, the critical path is the path
with the longest task time estimates. The critical path actually represents the fastest amount of
time the project can be finished it, as all the tasks in the critical path must be finished, but it is
assumed that shorter paths can be done within that same time. The critical path is therefore the
minimum amount of time a project can take, assuming there are enough resources that other

Agile Planning for Software Products | 46

paths can be completed at the same time. Let us examine the previous example to illustrate
this.

The critical path from preparing ingredients through to adding meatballs to the dish
(1268112) takes a total of 38 minutes. This is the longest task time estimates, as
other paths take shorter to finish. 1591112 takes 10 minutes, 14791112
takes 18 minutes, 181112 takes 32 minutes, and 131012 takes 5 minutes. Even
though the critical path takes the longest, all of those steps must be finished, so the overall
project will take at minimum 38 minutes to finish.

Also as in CPM charts, paths other than the critical path have slack, as they have some room
for estimates to change (as long as the path still takes less time than the critical path), without
affecting the overall project estimate.

Again, drawing from the above example, the paths 1591112, 14791112,
181112, and 131012, which are all shorter than the critical path, can take some
extra time to finish, if necessary, as long as they do not take more time than a total of 38
minutes for the path (or the overall project estimate will change). This flexibility in timing means
they have slack.

Any task on the critical path that is extended in terms of time will result in an increase to the
minimum time for the project, because there is no slack on the critical path.

In the above example, the critical path of 1268112 takes 38 minutes, so the minimum
time the project will take is 38 minutes. However, if the original estimate of 6 minutes to finish
task 68 is extended to 10 minutes, then the critical path estimate will also change, from 38
minutes to 42 minutes. The minimum time for the project has changed to a total of 42 minutes,
as the critical path has no slack.

In a PERT chart, critical paths must be horizontal. In the example above for the project of
making a spaghetti and meatball dish with a side salad, the critical path has been outlined in red
to be easily identified. As critical paths are always horizontal in PERT charts, they are always
easily identified, even if information such as time estimates may not be available. All other paths
in the chart must be drawn either above or below the critical path.

CPM Chart or PERT Chart?

In what circumstances is it better to use a CPM chart or a PERT chart? Both can be used to
demonstrate project dependencies and to generate time estimates. In other words, both work
well for task planning.

In some cases, either chart can be used with success, and it is then a matter of preference for
the development team and the project manager. In circumstances where the project may be
more task focused, however, the CPM chart is favourable. In CPM charts, additional information
can be added to tasks, such as costs. In circumstances where the project may be more event or
milestone based, in contrast, the PERT chart is favourable. PERT charts show what tasks need
to be completed prior to an event or milestone.

Agile Planning for Software Products | 47

Iteration Plans
As explained in the Release Plans lesson in Module 2: Estimation of this course, planning
occurs in both an iteration and a release level. Iteration planning involves dealing with tasks on
a detailed level, while release planning deals with broader level user stories.

Release plans therefore plan a few sprints at a time by selecting user stories from the project
backlog to be completed and released in a sprint. User stories are chosen for sprints by their
priority level. Release planning occurs before iteration planning because it happens before
sprints begin.

Iteration plans, on the other hand, are used to manage tasks within a sprint itself. They are
based on velocities, task estimates, and available time. Iteration plans ensure that a
development team has not over- or under-committed in each iteration. This keeps developers
focused on their tasks, makes certain the right tasks are finished, and that the project is on
schedule. Iteration plans are key for effective project development and maintaining a
sustainable project pace.

Most Agile methodologies, including Scrum and Extreme programming (XP), make use of
iteration plans. This lesson uses Scrum terminology such as sprints, product owner, and scrum
master to explain iteration plans. For more information on these terms, see the Software
Processes and Agile Practices course. The terms iteration and sprints are used
synonymously in this lesson as well.

Iteration plans are created in Sprint Planning meetings. Sprint Planning TIP: Four hours is
meetings have fixed lengths, set at the beginning of the meeting, and considered a
are therefore time-boxed meetings (see the Time Boxing lesson in good amount of
Module 2: Estimation of this course). As a time-boxed meeting, the time for a Sprint
length of Sprint Planning meetings must be strictly adhered to so that Planning meeting.
the meetings themselves stay on topic and that the iteration plans are
not over-planned.

Creating an Iteration Plan

The steps involved in creating an Iteration Plan during a Sprint Planning meeting are:

1. Create a Sprint Goal at the beginning of the Spring Planning meeting, if one does not
already exist for this sprint within the release plan.

The Sprint Goal is the general focus for the sprint being planned. Outlining the sprint
goal helps keep the development team focused on relevant tasks throughout the sprint.

2. Report the project velocity from the last sprint. This is an important step, as the previous
project velocity will determine the commitments made for the sprint being planned.

In cases where the last sprint was abnormal, the lowest velocity from previous sprints
should be used.

3. Determine all potential user stories to be completed within the sprint, and break them
down into developer tasks.

Agile Planning for Software Products | 48

 User stories selected for completion within the sprint will be informed by the release
plan. However, user stories from the previous sprint that were not considered done (for
example, if a story did not pass acceptance tests) should also be included in the iteration
plan for this sprint.

The number of user stories should be chosen based on the previous sprints velocity.
This means that there should only be enough stories within the sprint being planned, so
the sum of their story point estimates is less than or equal to the previous velocity. For
example, if the previous sprint finished 15 story points, then user stories should be
selected for this sprint until they equal 15 story points.

User stories are also generally chosen based on prioritythe most important ones
should be finished first. Any user stories not chosen must wait for a future sprint.

Once user stories have been chosen for the sprint,

they must then be broken down into developer tasks Kanban Boards, first
by the development team. Clients are not involved discussed in the course
with this task breakdown. A developer task should Software Processes &
only take a day or two to complete. Agile Practices of this
specialization, are a
After tasks have been broken down, they should be scheduling s ystem
developed by Toyota. This
expressed using techniques such as a Work system organizes
Breakdown Structure (WBS), a Critical Path Method production in a logical
(CPM) chart, a Program Evaluation and Review chain.
Technique (PERT) chart, a Kanban Board, etc.

4. Developers must create an estimate for each task.

Any task estimates must be agreed upon by the entire development team. In Scrum,
tasks are considered cross-functional. This means that any developer should be able to
take on any task. However, every developer has a unique pace, so their estimates may
differ for the same task. Task estimates are therefore an agreed average across the
team. Even tasks that require specialized skills, and therefore needs to be done by a
specific person on the development team, must have an estimate that the entire team
agrees upon.

Estimates are given in hours, half-days, or days. Any task with a smaller estimate than
these measures should be grouped with other tasks. Likewise, any task estimated to
take longer than three days should be broken down further.
5. Revisit chosen user stories for the sprint after estimates have been made. Once tasks
have estimates, it is easier to ensure they can be realistically done within the sprint.

Upon revisiting user stories, if it is found that

Estimates made at the iteration level they cannot all be finished, then the
override estimates made at the commitment for the sprint needs to be adjusted.
release level. Iteration level User stories and their associated tasks should
estimates are more likely to be be removed from the iteration plan until the total
accurate, because iteration plans are
estimated time matches the available time of
more specific. If iteration level
estimates change release level
the developers.
estimates, the release plan should be
changed accordingly. Typically,
release plans often only look at a few Agile Planning for Software Products | 49
sprints, and are regularly adjusted.
 On the other hand, if it is found that the selected user stories can all be finished within
the sprint with time left, then more user stories can be added, as long as the total time
doesnt go over the time of the sprint.

6. The development team must sign up for tasks.

In iteration plans, tasks are self-assigned. This means that managers do not assign
tasks to members of the development team, but rather, developers choose to work on
what they find interesting. This generally results in more satisfied workers, and better
software.

Developers should choose tasks based on their available time, so other commitments or
projects the developer is involved in can be accounted for.

Iteration plans allow both the development team and the client to know what to expect from a
sprint and what the product at the end of the sprint will look like. The team and the client should
be on the same page.

Sample Iteration Plan

SPRINT GOAL: In this sprint, the team is creating the user profile elements. At the completion
of this sprint, users should be able to create accounts, log in, and edit their profile.

LAST SPRINT: In the last sprint, we completed 33 story points. It is based on three developers
working approximately 40 hours per week for the two-week sprint.

Agile Planning for Software Products | 50

 NAME/ PRIORIT SIZE ASSIGNED EFFORT ESTIMATE (DAYS)
DESCRIPTION Y ESTIMATE TO
(POINTS)

CREATE 1 2 DAVID 3
DATABASE FOR
USER
ACCOUNTS

DESIGN LOGIN 1 2 DAVID 1

PAGE

DESIGN 1 1 CLAIRE 1
ACCOUNT
CREATION PAGE

DESIGN USER 1 1 ALEX 1

PROFILE PAGE

PROGRAM 3 1 DAVID 2
DATABASE
FUNCTIONALITY
TO LOGIN PAGE

PROGRAM 3 1 CLAIRE 2
DATABASE
FUNCTIONALITY
TO ACCOUNT
CREATION PAGE

PROGRAM 2 1 DAVID 2
LOGIN PAGE

PROGRAM 2 1 CLAIRE 2
ACCOUNT
CREATION PAGE

PROGRAM 2 2 ALEX 2
PROFILE PAGE

WRITE TESTS 2 1 DAVID 2

FOR ACCOUNT
LOGIN

WRITE TESTS 2 1 CLAIRE 2

FOR ACCOUNT

Agile Planning for Software Products | 51

CREATION

WRITE TESTS 2 1 ALEX 2

FOR USER
CHANGING
PROFILE
PICTURE

WRITE TESTS 2 1 ALEX 2

FOR USER
CHANGING
INFORMATION

RUN 4 2 DAVID 2
ACCEPTANCE
TEST ON LOGIN
PAGE

RUN 4 2 CLAIRE 2
ACCEPTANCE
TEST ON
ACCOUNT
CREATION PAGE

RUN 4 2 ALEX 2
ACCEPTANCE
TEST ON
PROFILE PAGE

WRITE USER 4 1 DAVID 1

MANUAL
DESCRIPTION
FOR LOGIN
PAGE

WRITE USER 4 1 CLAIRE 1

MANUAL
DESCRIPTION
FOR CREATING
AN ACCOUNT

WRITE USER 4 1 ALEX 1

MANUAL
DESCRIPTION
FOR CHANGING
PROFILE

TOTAL 25 33

Agile Planning for Software Products | 52

Module 4: Risks

Upon completion of this module, you should be able to:

Identify, describe, and suggest means of addressing common management anti-patterns,

including: analysis paralysis, cart before the horse, groupthink, silos, vendor lock-in,
overengineering, gold-plating, viewgraph engineering, fire drill and heroics, death march,
micromanagement, seagull management, email as the primary means of communication, loose
cannon, and intellectual violence.
Understand the difference between group anti-patterns and individual anti-patterns.
Define the term risk.
Identify and define types of risks including: scope risk, technology risk, customer and
stakeholder risks, and personnel risks.
Define the terms impact, likelihood, and value.
Know how to create and use an impact vs. likelihood matrix and a risk-value matrix to prioritize
risks and features.
Know how to address different rankings of problems along the impact vs. likelihood matrix and
risk-value matrix.
Define the terms risk management plan, indicator, action.
Create a risk management plan.
Understand that some risks are unforeseeable and must be dealt with as they come.

This module focuses on risks and risk planning. Risks can make even a well-planned Agile
project go off track, so it is important to plan for and mitigate risks as much as possible.

Anti-Patterns
Some risks are so common, they create predictable patterns in projects. These risks are known
as anti-patterns. Anti-patterns are defined as commonly occurring solutions or situations in a
project that cause negative consequences such as project failure. They are the classics of risk
management and are not project-specific risks.

The best way to avoid anti-patterns is to understand them, so they can be identified and
addressed.

Many different types of anti-patterns exist, including:

Anti-patterns from writing code
Software architecture anti-patterns
Management anti-patterns

This module focuses on management anti-patterns. This type of anti-pattern can have many
different names, but the concept remains the same: management anti-patterns are ways in
which projects fail due to the behaviour of individuals or groups of people.

Only a few anti-patterns are listed in this lesson. For a full list, and accompanying links, see
Course Resources.

Agile Planning for Software Products | 53

Group Anti-Patterns
Group anti-patterns refer to risks involving groups of people. The following group anti-patterns
are covered in this module:

Analysis paralysis Over-engineering

Cart before the horse Gold plating
Groupthink Viewgraph engineering
Silos Fire drill and heroics
Vendor lock-in Death march

Analysis Paralysis
Analysis Paralysis is a group anti-pattern where the development team gets stuck or stalled in
some phase of the project, usually from analyzing requirements, leading to paralysis of the
project.

This commonly happens in the specification phase of the project. It is during this phase that
projects are likely delayed because clients and/or product managers spend a lot of time
analyzing requirements and cannot decide on a direction for the project until the analysis is
perfected.

Although wanting to avoid choosing the wrong path is admirable, spending long periods of time
waiting for information is not in the spirit of Agile methodology. Instead, progress should be at a
constant pace in a projectso as to avoid unhappy developers.

To avoid this anti-pattern, a good strategy is run a project with incremental releases, as
embraced in Agile. With incremental releases, not everything needs to be known upfront
because it is understood the product must be flexible and will change and refine over time.

In the course Software Processes and Agile Practices, you learned about the Agile practice of Lean
Software Design. One of the seven principles of Lean is:

Deliver as fast as possible

This principle means the product should evolve through a series of rapid iterations. This is
advantageous not only for avoiding analysis paralysis, but because it allows the client to give feedback
and it places focus on core product features.

Cart Before the Horse

Cart before the horse is also a group anti-pattern. The name of this anti-pattern comes from
the English idiom, putting the cart before the horse. This phrase means that something was
put in a counter-productive order, just like putting a cart in front a horse is less productive than
having the horse pull a cart. In software product management, cart before the horse similarly
means putting too much emphasis on a part of the project that should be done later.

Agile Planning for Software Products | 54

 In the course Software Processes and
Agile Practices, you learned about the
Agile practice of Lean Software Design.
Lean Software Design solves the issue of
groupthink by allowing individuals to come
up with solutions to designs and other This anti-pattern might be easier understood
problems on paper, without talking. This isthrough an example. If less time and resources are
what generating solutions silently means.spent developing a critical feature of a product
Then, in a group, each persons idea is than on a less critical one, or a feature that is not
discussed in turn, and a final solution required at the moment, this is an example of cart
combining the entire groups designs is before the horse. This frequently happens when
created. programmers assign themselves work that doesnt
directly impact current features, in order to get a
head start on future work. However, the danger of this process is that a team can find
themselves with a lot of poorly developed features of varying importance.

To avoid this anti-pattern, it is important for the development team to understand development
priorities and stay focused on those priorities. In other words, the team should clearly
understand what work is to be done now and what work is to be done later.

Groupthink
Groupthink is a group anti-pattern that refers to how people tend to follow the general opinions
of a group, regardless of whether or not their own individual opinions are different. The term
comes from the social sciences. Groupthink can lead to poor decisions and judgments in
project development.

The most common example of groupthink is in group

discussion on projects, when ideas tend to come from
In the course Software Processes and
only one or two people in a meeting, while most other Agile Practices, you learned about the
members are quiet and go along with suggestions. Agile practice of Lean Software Design.
This can lead to a poor end product, because often One of the seven principles of Lean is:
other or better options are not discussed.
Eliminate waste
To avoid groupthink, there are a couple of good
measures to take. First, it is important to remember This principle means that it is important for
that conflict can be a good thing in group discussions, the development team to be careful of
if done constructively. Creating an open atmosphere being busy instead of being productive. A
busy team can get distracted by extra
of discussion, where different perspectives are
features, which can ultimately interfere with
encouraged, is very helpful in combating groupthink. the functionality of the end product, rather
than focus on developing the core
Another means of addressing groupthink is to requirements. This is wasteful and should
generate solutions to problems silently, as in Lean be avoided.
Software Design. This means allowing a team to
break up into smaller groups or individuals and encouraging these smaller groups to come up
with different designs and alternatives. These suggestions are then combined together. This
approach has the benefit of not putting people on the spot.

For a video demonstration of groupthink in action, see the Course Resources.

Silos
Silos are group anti-patterns that occur in the extreme opposite of groupthink. A team
separated in smaller groups can be a good strategy, particularly if specialized groups create

Agile Planning for Software Products | 55

more effective work, but this then runs the risk of teams losing touch with each other and limiting
communication. This is called working in silos.

In essence, silos are created when a lack of communication among groups on the development
team occurs, which lead to a loss of unified focus and the creation of counterproductive
strategies and product features. Features developed in silos are developed in a vacuum, so the
work of one team could be different or incompatible with that of another team, and bringing the
work of two different teams together could thus prove difficult.

To address this problem, it is important to encourage an open and positive atmosphere where
team communication and collaboration is favoured. This also encourages the development team
to be less apathetic. Alternately, re-examining management structure could also be beneficial. If
there is flat management where developers are able to speak directly to other developers
instead of going through managers or the managers of other teams, communication can flow
easier and bureaucracy can be avoided. In both cases, face-to-face communication should be
encouraged.

Vendor Lock-in
Vendor lock-in is a group anti-pattern that occurs when a development team creates a product
that relies heavily on a single technology solution or vendor. Vendor lock-in is different from
choosing a technology because it is the best option, but rather it generally involves a lack of
flexibility in moving to another technology without substantial costs.

The heavy reliance on a single technology as a project moves forward is problematic if the
technology cannot cover what is needed, becomes outdated, or cannot adapt to change. In
extreme cases, staff can even leave an organization over vendor lock-in.

To avoid this risk, it is important to do research before committing to a technology or a solution,

regardless of recommendations made to the project.

Over-engineering
Over-engineering is a group anti-pattern that relates to how the development team creates the
product itself. It occurs when a product is made more complex than necessary, such as when
extra or needless features are added to a product, making it confusing for the majority of users.
Over-engineering can happen on both the user interface (UI) of a product or in its internal
processes.

There are many examples of over-engineering, such as digital cameras with too much features
that must be set before functionality is enabled, music players with too many options, cars that
can achieve speeds that it will never run at, or text editors with too many save options.

To avoid over-engineering, it is important that the development team clearly understands the
needs of a product (what the product needs to be able to do) versus the wants of a product
(what would be nice if the product could do but is not necessary for success). In the course
Client Needs and Software Requirements, determining needs and wants was discussed at
length, including the importance of prioritizing requirements and ensuring a project is feasible.

Agile Planning for Software Products | 56

Gold Plating
Gold plating is a group anti-pattern that relates to how a development team creates a product,
much like over-engineering. It occurs when so much effort is put into one part of a project that it
reaches a point of diminishing returns. In other words, so much extra effort is focused on a part
of the project that it no longer contributes value to the product.

This usually occurs when the development team adds extra features to a product in order to
impress the client, particularly when the team finishes early. However, added features can add
unforeseen, extra work to a project, and features that the client did not actually want in the
product (remember that user requirements should be specified by the client and not by the
development team). This can make the client disappointed and confused in the end product.

To avoid gold plating, the development team should stop working when the product functions
well. If the team feels new features should be added, the features must first be vetted with the
client. Good questions to ask when considering adding extra features include:

Does this feature improve the product?

Does this feature make the product confusing somehow?
Does this feature take away from main functionality?

Another solution to avoiding gold plating could be conducting a user study. A user study
involves taking a small sample of users and asking them to try out new features on a product
and provide feedback. This feedback can inform the team on how useful the feature is, how it
affects the interface, whether or not the feature has broad appeal, and how the product itself
can be improved. Not all extra features are bad things, but these steps help to ensure that the
extra features actually add value to the product.

Viewgraph Engineering
Viewgraph engineering is a group anti-pattern that occurs when too little effort is put into a
project. It is the opposite of gold plating in that way. However, if gold plating is unfocused work,
viewgraph engineering is working on what is not important.

Viewgraph engineering is not the same as hiring bad developers and missing deadlines, which
is also a risk but a separate issue. Instead, viewgraph engineering is more directly tied to
management practices and how those practices can make project work difficult for developers
by requiring them to work on things other than development work, such as documentation,
reports, or creating presentations. The time spent on such activities should not be greater than
the time spent writing code or developing a project.

This anti-pattern commonly occurs in organizations with vertical management structures, where
developers must constantly analyze the potential of a concept and prove it to management
before going forward. In this way, viewgraph engineering is a kind of analysis paralysis on an
organizational level.

Agile methods purposefully seek to avoid viewgraph engineering by placing focus on creating
working software over comprehensive documentation. Although documentation should still be
produced, it should not lead to unfocused development work.

Agile Planning for Software Products | 57

In a similar vein, the best solution to viewgraph engineering is to remove any barriers
developers might face in developing a project. Creating a basic prototype is often more
informative than any report or presentation, so time spent on such documentation may be better
served in development. If the prototype is good, then it can be built upon. If it is not good, then it
can be scrapped, and the team will learn for going forward in both the project, and future
projects.

Fire Drills (Heroics)

Fire drills and heroics are two types of linked grouped anti-patterns. Both are very common.

Fire drills happen when the development team does very little work throughout most of the
project and then makes a big push towards the end of the project, resulting in overtime work in
order to meet deadlines. This leads to a loss of quality in favour of producing a product quickly.
Fire drills can have many causes, including:

Bad management practices

Developers wasting time
Expectations were not properly established with the client

Fire drills can also lead to heroics. Heroics is an anti-pattern where the development team ends
up relying on only one persons technical skills to get the project done. This requires that one
developer to have almost superhuman skills, as he or she take on almost the entire project.
Further, the team relies on only one person, which is bad for both the team and the developer.

To avoid fire drills and heroics, it is important to establish expectations with the client early on
the project and to follow Agile software development practices. For example, using time boxes
and always producing a working product at the end of those time boxes. These strategies keep
the team working at a stead pace, avoiding both fire drills and heroics. They will also keep up
team motivation for the project.

Death March
Death march is a group anti-pattern that happens when the development team and the
management team are at odds, because the management team is enforcing a path for the
project that the development team does not believe in. This leads the development team to lose
conviction and passion for the project, which increases likelihood of project failure. In spite of
this, everyone keeps working out of a sense of obligation.

Death marches can occur because of financial reasons, or when management is too stubborn to
recognize other ideas and the potential failure of the project. In either case, if developer morale
is low, then the quality of the product will suffer.

To avoid death marches, it is important for management to maintain open communication with
the development team. In other words, managers should listen to what the team thinks about
the direction of the project and be willing to explore other alternatives in project development. In
some cases, simply reassuring the development team that you understand and recognize their
concerns is all that is needed to avoid a death march.

Agile Planning for Software Products | 58

Individual Anti-Patterns
Individual anti-patterns are anti-patterns caused by individuals instead of group behaviour.
Individuals caught up in these anti-patterns usually find it hard to recognize it, but people
outside the situation can identify these anti-patterns easily.
In general, work cultures that promote open mindedness, self-reflection, and improvement
greatly help to counter individual anti-patterns.

The following individual anti-patterns are covered in this lesson:

Micromanagement
Seagull management
Email as the primary means of communication
Loose cannons
Intellectual violence

The first two anti-patterns listed arise from the actions of individual managers, while the last two
come from the actions of individual team members.

Micromanagement
Micromanagement is a very common individual anti-pattern. It occurs when a manager is very
controlling and wants to be involved in every detail of a project, no matter how small. The
manager needs to constantly know what their developers are doing. It demonstrates a lack of
trust in the development team, which affects team morale and product quality. This can become
exacerbated if the team is blamed for the managers behaviour.

Micromanagers, in general, do not micromanage with the intention of bringing down developer
morale. Instead, they usually think that a project will fall apart without them. Micromanagement
is therefore usually born out of a managers own internal fears, insecurities, or basic stresses. It
can also be caused by overly ambitious timelines, poor product quality, or fear that developers
are not up to the job.

Addressing micromanagement can be difficult, as there is no quick and easy fix. In general,
micromanagers themselves have to provide a solution. Micromanager has to admit their
behaviour is affecting the team and be willing to make steps towards improving their behaviour.
Further management training may be needed before micromanagers can adjust their actions.

It is worth noting that wanting to stay informed on project progress should not be confused with
micromanagementin fact, asking for short daily meetings updating project progress is good
Agile practice!

Seagull Management
Seagull management is an individual anti-pattern that happens when a manager is only
present when problems arise or right before deadlines. This results in an otherwise absent
manager swooping in, dumping a lot of information or requests on the development team, and
then disappearing again. This can cause a lot of stress for the team and can easily push a
project into become a fire drill project. Further, the development team will be working in constant
fear of when the next dump will be.

Agile Planning for Software Products | 59

The behaviour of seagull managers is usually caused by a number of factors, such as busy
schedules, inadequate planning, or stress. Addressing seagull management behaviour involves
allowing a manager to lighten their workload (although this might not always be a solution that
product managers can employ, because it depends on things outside the scope of the
managers themselves). In general, if you suspect you might be a seagull manager, it is a good
idea to re-evaluate your schedule and how you interact with the development team.

Email as the Primary Means of Communication

Email as the primary means of communication is such a common issue in projects that it has
become its own anti-pattern. The methods of communication used between managers and
development teams can greatly impact a project. Email, in particular, lends itself to creating
seagull managers who only write to the team with large workloads or big dumps of information,
and do not communicate frequently otherwise (see above). Further, email is often not the best
way for the team to communicate well with managers because it can be difficult to provide
context in emails, or alternately, information can be easily misunderstood. Emails can also be
too formal for most communications.

To address this anti-pattern, it is important that managers and development teams take time to
communicate in person. When that is not possible, other means of communication such as chat
services, video conferencing, and phone calls are good alternatives to email. All of these forms
of communication have the added benefit of giving respect and attention to team members,
which help to increase team motivation.

Loose Cannon
A loose cannon is an individual anti-pattern. The term loose cannon itself refers to an
unpredictable person (or thing) who can negatively affect others, unless it is managed
somehow. In software projects, loose cannons tend to be people who make significant project
decisions without consulting the rest of the team. This behaviour is reckless and can create
more work for other group members.

A loose cannon is not to be confused MORE INFORMATION

with someone who asserts their opinions
on every topic, leading to poor project The origin of the term loose cannon comes from
decisions. This is more closely related to the 17th and 19th centuries when wooden sailing
ships used cannons as weapons. When a cannon is
the anti-pattern of intellectual violence,
fired, it undergoes a large amount of recoil. To avoid
discussed below. damage from this recoil, cannons were tightly
Like many individual anti-patterns, there secured to decks using ropes.
is no quick fix to loose cannons. Often,
dealing with loose cannons involves A loose cannon refers to a cannon that has come
understanding the motivations of the free of these restraints, allowing it to roll around the
individuals behaviour. Are they trying to ship and cause damage. The term has survived until
cause trouble? Are they trying to prove this day!
their work to the team?

Once this is understood, a manager or team can try and get the individual to understand that
their behaviour is destructive and encourage the individual to make steps to change it. In some
cases, organizational steps might be necessary to address the problem if the personality of the

Agile Planning for Software Products | 60

loose cannon makes change very difficult. In any case, loose cannons should be dealt with
quickly or the risk to a project will increase over time.

Intellectual Violence
Intellectual violence is an individual anti-pattern that concerns a person who asserts their
opinion on every topic and impedes progress by questioning every decision and action or using
their advanced knowledge in an area to intimidate or look down on other team members. These
individuals tend to repeat their opinions so much that the rest of the group concedes to them
just to avoid confrontation. In turn, this can lead to low team morale, with bitterness and apathy
among group members that affects open communication and productivity. Loose cannons are
sometimes the cause of intellectual violence.

There are a couple of strategies for addressing intellectual violence. The project manager can
try talking to the individual in private about his or her behaviour and suggest any appropriate
changes. Alternately, the project manager can encourage an open question policy (for example,
reinforcing the idea that there is no such thing as a stupid question), and discourage prejudging
opinions or inexperience. Intellectual violence should be addressed quickly to prevent a project
from taking a bad direction and to keep up team morale.

Causes of Failures
Projects can fail not only from issues that present themselves as patterns, but also from project
risks. Risk is something that could potentially cause a project to fail. Risks are found in every
situation, not just in projects and project plans, so it is important to try and plan for them.

There are many different types of risks. These include:

Scope risks, which refer to risks that involve expanding requirements. Agile methods
are well equipped to address scope risks, because Agile has good change control
measures. In other words, Agile deals well with change and changing requirements.
Technology risks, which refer to risks that the technology used in a project will fail. This
includes hardware failure, software protocols becoming obsolete or unsupported, lack of
scalability with technology, or a lack of technological understanding on the part of the
development team.
Customer and stakeholder risks, which refers to risks that customers and stakeholders
can bring to a project. These usually occur when customers promise to provide material
or information to the team and they forget or deliver it late. Examples of customer or
stakeholder risks include clients becoming apathetic or a change in the main point of
contact between the team and the client.
Personnel risks, which refer to risks posed by personnel on the development team. For
example, team members may leave partway through a project. This can be especially
problematic if the person who leaves has specialized skills that are not shared by other
team members. Personnel risks could also refer to a lack of needed skills, conflict
between team members, or communication issues, such as groupthink.
Many other risks, such as legal issues, security problems, the destruction of physical
locations, loss of interest from either management or development teams, industrial
espionage, etc.

Agile Planning for Software Products | 61

Risk Assessment, Likelihood, and Impact
In order to plan for risks, it is important not only to be able to identify potential risks, but to
understand both the likelihood of a risk happening over the course of a project and the impact of
that risk on the project.

Risks may include the anti-patterns as discussed above as well as project-specific problems
that may not be common. In order to assess risks, their likelihoods, and impacts, the first step is
to create a list of risks. You should recognize that you can take action to avoid some risks, but
this is not always possible.

Once risks have been identified, they can be prioritized. One of the most useful ways of
prioritizing risks is using an impact vs. likelihood matrix.

Impact vs. Likelihood Matrix

An impact vs. likelihood matrix is a two-dimensional representation of the amount of impact a
risk has on the project and how likely that impact is to happen. With this knowledge, risks can
prioritized, and the team will know what to focus on preventing, and what to monitor.

In an impact vs. likelihood matrix, impact runs along the y-axis, from low to high. Likelihood runs
along the x-axis, also from high to low. Different grids in the matrix are assigned different
rankings based on how high the impact and likelihood is.

To use the matrix, the listed risks of the project should be assigned a value from the matrix
based on their value along each axis. For example, if your project has identified the risk project
runs out of funding, this is potentially a big risk. It is high on impact. But, if this only has a
medium likelihood of occurring, then it would go in the middle column of the matrix, at the top of
the grid.

Risks can range quite broadly, from being an immediate impact to something with a very slim
likelihood, like asteroid destroys the Earth.

Agile Planning for Software Products | 62

The numbers in each cell of the matrix represents a different risk ranking, as determined by
each risks value on the matrix. For the examples above, project runs out of funding has a risk
ranking of 2, as seen in middle column of the top row. The risk asteroid destroys the Earth is a
large impact, so it is in the top row. However, it also has a low likelihood, so it will be in the third
column. It will therefore be situated in the bottom-right square, and has a risk ranking of 3.

After a list of risks has been made and assigned rankings, the team is informed of the potential
magnitude of risks the project faces, so appropriate action can be taken regarding the risks.

Based on the ranking of the risk, the following actions are recommended:
If there is a ranking of 1, then the team needs to mitigate the damage. The risk needs to
be addressed as thoroughly as possible and quickly, as it is the most dangerous.
If there is a ranking of 2, then the team should be concerned about the risk but do not
need to take immediate action, since it is only moderately likely to happen. A strategy
should be developed, however, for dealing with the risk should it come to pass.
If there is a ranking of 3, then the team should spend very little time considering the risk,
as it is not of great concern to the project and it is the least threatening.

Other variations on the impact vs. likelihood matrix are possible. Some projects may assign risk
rankings differently. For example, some projects may assign a ranking of 1 to the high
impact/high likelihood cell in the top left, as above, but also to the cells of high impact/medium
likelihood and medium impact/high likelihood. In this scenario, a ranking of 2 is assigned to cells
on the edges of those cells with a ranking of 1, and 3 to everything else. This puts a higher
importance on some risks, which is known as decreasing a projects risk tolerance. This practice
is important for mission-critical projectsprojects that are important to a business or
company, because they are tied to the success of that business or company.

Any actions taken to deal with risks adds activities to the project plan. These activities should be
taken into account because they can influence both project deadlines and costs.

Risk-Value Matrix
A matrix is a useful tool that can also be applied to project features. This kind of matrix is known
as a risk-value matrix. It is similar to an impact vs. likelihood matrix, but it focuses on showing
how likely a risk is to influence the value of a project feature.

Accordingly, the y-axis of a risk-value matrix shows risk, while the x-axis shows the value of the
feature to the project. Instead of assessing risks with the matrix, features are assessed. This
helps prioritize which features the development team should work on first.
For example, consider the hypothetic project of building a mobile app with a database to store
data. Lets say the development team is very experienced in creating databases but has little
experience creating mobile apps. In this case, on a risk-value matrix, database features are low
risk and high value, while app features are high risk and high value.

Using the ranking information the risk-value matrix provided, development priority can be
assignedthose features that are high risk and high value should always be prioritized first. In
this case, developing app features should be done first. Although instinct suggests that it is
better to start with developing high value and low risk features first, as it theoretically be done
with more ease, this is wrong. Tasks with high risk pose the most danger to success, and
problems associated with these tasks should be tackled quickly. This way, any issues that may

Agile Planning for Software Products | 63

cause delay or cancellation can be identified earlier in the project, rather than later, and time
and work is not lost working on other features.

Risk Strategies, Contingency, Mitigation

In addition to be able to identify risks, likelihood, impact, and effects on project feature values, it
important to be able to create a risk plan and implement it.

A risk plan, or a risk management plan, is a list of potential project risks, associated impacts,
and likelihood, and any planned action for how to address the risk should it occur. Generally,
risk management plans are presented in tables.

The table consists of four columns. From left to right, these columns contain:
The risk
The associated rank of that risk, as determined by the impact vs. likelihood matrix
A description of the risks indicators (An indicator is a sign that the risk is going to occur
and is usually some sort of event or activity to look out for.)
A description of the action to take if the risk presents itself

A risk management plan is better understood through an example. Imagine a project that seeks
to create a music player app. In order to have a functional app, the client needs to obtain
copyright access to the content they wish to provide users.

In an impact vs. likelihood matrix, it is determined that the risk unable to obtain copyright to
music for app has a ranking of 2. It is high impact but only moderately likely to happen.
Indicators that a failure to obtain copyright has occurred include clients or musicians
communicating this to the team or if clients or musicians lose interest in the app. A number of
actions can be taken to address this issue, such as:

Ensuring the development team delivers the product in increments, so the client can see
progress in the product and remain actively involved and interested in it.
Settling financials of the project appropriately, so there is a guarantee that work done will
be compensated.
Some product managers may get involved with helping the client obtain copyright. This
level of involvement largely depends on the scope of the individual position of the
product manager.

In a risk management plan table, it would look like:

Risk Risk Indicators Action

Ranking

Agile Planning for Software Products | 64

 Unable to obtain 2 Musicians or client Deliver project in
copyright to music for communicate increments so progress
app difficulties in obtaining is visible to client
copyright Settle financials ahead
Musicians or client of time so there is no
lose interest in app risk of not being paid
for work
Help client gain
copyright access

Using a risk management plan for all risks ranked as 1 or 2 will greatly increase the potential
success of the project.

Even with risk management plans, it is very important that the product manager and
development team recognize that there will always be risks that cannot be predicted or planned
for, as every project will face unique risks. However, this does not mean that risk planning
should be ignored. Product managers should recognize that dealing with the unknown is a skill
they will need to develop.

In cases of unforeseen risks, project success will largely depend on how those risks are dealt
with in the moment.

Agile Planning for Software Products | 65

 Copyright 2016 University of Alberta.

All material in this course, unless otherwise noted, has been developed by and is the property
of the University of Alberta. The university has attempted to ensure that all copyright has been
obtained. If you believe that something is in error or has been omitted, please contact us.

Reproduction of this material in whole or in part is acceptable, provided all University of Alberta
logos and brand markings remain as they appear in the original work.
Version 0.7.2

Agile Planning for Software Products | 66