An Agile Approach to Estimating and Planning
Mike Cohn is a co-signer on the Addison-Wesley Signature Series and the author of Succeeding with Agile: Software Development Using Scrum (December 2009).
“A good plan violently executed now is better than a perfect plan executed next week.”
—General George S. Patton
Although it started years before, the agile movement officially began with the creation of the Agile Manifesto in February 2001 (Beck et al.). This manifesto was written and signed by seventeen “lightweight methodologists,” as they were called at the time. Their document both gave a name to how they were developing software and provided a list of value statements. The authors of the Agile Manifesto wrote that they value
- Individuals and interactions over processes and tools
- Working software over comprehensive documentation
- Customer collaboration over contract negotiation
- Responding to change over following a plan
Agile teams value individuals and interactions over processes and tools because they know that a well-functioning team of great individuals with mediocre tools will always outperform a dysfunctional team of mediocre individuals with great tools and processes. Great software is made by great individuals, and as an industry we have tried too long with too little success to define a development process that relegates individuals to replaceable cogs in the machinery. Agile processes acknowledge the unique strengths (and weaknesses) of individuals and capitalize on these rather than attempting to make everyone homogeneous.
Agile teams value working software over comprehensive documentation because it leads them to have a stable, incrementally enhanced version of the product at the end of each iteration. This makes it possible to collect early, frequent feedback on both the product and the process. As the developed software grows each iteration, it can be shown to likely or actual users. Feedback from these users is fed back into the development process to make sure that the team is always working on the highest-valued features and that those features will satisfy user expectations.
Customer collaboration is valued over contract negotiation because agile teams would like all parties to the project to be working toward the same set of goals. Contract negotiation sometimes sets the development team and the project customer at odds right from the start. I enjoy playing most games, and when my oldest daughter was four, I bought her a “cooperative game” because it looked like a game she’d enjoy and because I had no idea how a cooperative game could be fun. In the game I bought her, a princess is placed under a spell, and players need to remove obstacles (a moat, a locked door, and so on) that are between them and the princess. Players take turns, as in most games, but the goal is to remove obstacles collaboratively and save the princess. All players win, or all players lose. The game is surprisingly fun, and we’d like software teams and customers to approach projects with this same attitude of collaboration and shared goals. Yes, contracts are often necessary but the terms and details in a contract can exert great influence on whether the different parties are set on a collaborative or a competitive effort.
Agile teams value responding to change over following a plan because their ultimate focus is on delivering as much value as possible to the project’s customer and users. For all but the simplest projects, it is impossible for users to know every detail of every feature they want. It is inevitable that users will come up with new ideas, and almost as inevitable that they will decide that some features desired today will become lower priorities tomorrow. To an agile team, a plan is one view of the future, but many views are possible. As a team gains knowledge and experience, they will factor these into the plan. Perhaps the team is progressing faster or slower than initially expected; perhaps users like one set of features more than expected but don’t like another feature that was initially considered critical.
With the four value statements of the Agile Manifesto in mind, in this chapter we consider what it means to have an agile approach to a project, as well as what it means to have an agile approach to estimating and planning.
An Agile Approach to Projects
With an understanding of the four primary agile value statements, we can turn our attention to what an agile team looks like in practice. Taken collectively, the four value statements lead to software development processes that are highly iterative and incremental and that deliver coded and tested software at the end of each iteration. The following sections cover some of the main ways in which agile teams work, including that they:
- Work as one team
- Work in short iterations
- Deliver something each iteration
- Focus on business priorities
- Inspect and adapt
An Agile Team Works As One
Critical to the success of a project is that all project participants view themselves as one team aimed at a common goal. There is no room for a “throw it over the wall” mentality on an agile project. Analysts do not throw requirements over the wall to designers. Designers and architects do not throw designs over a wall to coders; coders do not throw half-tested code over a wall to testers. A successful agile team must have a we’re-all-in-this-together mindset. Although an agile team should work together as one whole team, there are a number of specific roles on the team. It is worth identifying and clarifying those roles that play a part in agile estimating and planning.
The first role is the product owner. The primary duties of the product owner include making sure that all team members are pursuing a common vision for the project, establishing priorities so that the highest-valued functionality is always being worked on, and making decisions that lead to a good return on the investment in the project. In commercial software development, the product owner is often someone from the marketing or product management side of the company. When developing software for internal use, the product owner may instead be a user, the users’ manager, an analyst, or the person funding the project.
A second role is that of customer. The customer is the person who has made the decision to fund the project or to buy the software. On a project developing software for internal use, the customer is usually a representative from another group or division. On such projects, the product owner and customer roles are often combined. For a commercially distributed product, the customer will be the person who buys the software. In either case, the customer may or may not be a user of the software, which is, of course, another important role.
Another role worth highlighting is that of developer. I use developer very generally to refer to anyone developing software. That includes programmers, testers, analysts, database engineers, usability experts, technical writers, architects, designers, and so on. Using this definition, even the product owner may be thought of as a developer on many projects.
A final role is the project manager. As described by Highsmith (2004a), the role of the project manager changes on agile projects. Agile project managers focus more on leadership than on management. On some agile projects, the person fufilling the role of project manager will also act in another role, often as a developer but occasionally as a product owner.
An Agile Team Works in Short Iterations
On an agile project there is no grand delineation of phases—no up-front requirements phase followed by analysis followed by architectural design and so on. Depending upon the actual agile process you select or define, you may put a very short design, modeling, or other phase at the front end of the project. But once the project has begun in earnest, all work (analysis, design, coding, testing, and so on) happens concurrently within each iteration.
Iterations are timeboxed, meaning they finish on time even if functionality is dropped. Timeboxes are often very short. Most agile teams work in iterations two to four weeks long, but some teams maintain their agility with iterations of up to three months. Most teams settle upon a relatively consistent iteration length. Some, however, choose the appropriate length for an iteration at the start of each iteration.
An Agile Team Delivers Something Each Iteration
More crucial than the specific iteration length chosen by a team is that during the iteration they transform one or more imprecise requirements statements into coded, tested, and potentially shippable software. Of course, many teams will not deliver the results of every iteration to their users; the goal is simply that they could. This means that teams make progress by adding one or more small features in each iteration but that each added feature is coded, tested, and of releaseable quality.
It is essential that the product be brought to this potentially shippable state by the end of each iteration. Practically, this does not mean a team must do absolutely everything necessary to release, because they often won’t release each iteration. For example, I work with one team that requires two months of mean time between failure (MTBF) testing before releasing their product, which includes both hardware and software. They cannot shorten those two months, as it is contractually required by their client, and that amount of time is often necessary to check for hardware failures. This team works in four-week iterations, and apart from running this two-month MTBF test, their product is at a truly releasable state at the end of each iteration.
Because a single iteration does not usually provide sufficient time to complete enough new functionality to satisfy user or customer desires, the broader concept of a release is introduced. A release comprises one or more (usually more) iterations that build upon one another to complete a set of related functionality. Although iterations are most commonly two to four weeks, a release is typically two to six months. For example, in an investment management system, one release may include all of the functionality related to buying and selling mutual funds and money market funds. This may take six two-week iterations to complete (roughly three months). A second release may add stock and bond trading and take four additional two-week iterations. Releases may occur at varying intervals. A first release may take six months to be developed. It may be followed by another release three months later, and so on.
An Agile Team Focuses on Business Priorities
Agile teams demonstrate a commitment to business priorities in two ways. First, they deliver features in the order specified by the product owner, who is expected to prioritize and combine features into a release that optimizes the return on the organization’s investment in the project. To achieve this, a release plan is created based on the team’s capabilities and a prioritized list of desired new features. For the product owner to have the most flexibility in prioritizing, features must be written so as to minimize the technical dependencies among them. It is difficult for a product owner to prioritize features into a release plan if the selection of one feature requires the prior development of three others. A team is unlikely to achieve a goal of absolutely no dependencies; however, keeping dependencies at a minimum is often quite feasible.
Second, agile teams focus on completing and delivering user-valued features rather than on completing isolated tasks (that eventually combine into a user-valued feature). One of the best ways to do this is to work with user stories, which are a lightweight technique for expressing software requirements (Cohn 2004). A user story is a brief description of functionality as viewed by a user or customer of the system. User stories are free-form, and there is no mandatory syntax. However, it can be useful to think of a story generally fitting this form: “As a <type of user>, I want <capability> so that <business value>.” With this template as an example, you may have the story “As a book buyer, I want to search for a book by ISBN so that I can find the right book quickly.”
User stories are lightweight because the work to gather and document them is not all done up front. Rather than writing a lengthy requirements specification, agile teams have found it better to pursue a just-in-time requirements approach. Typically this begins with a short description of a user story being handwritten on a note card or perhaps typed into a computer for larger or distributed teams. The story card is just the beginning, though, and each user story is accompanied by as many conversations between the developers and the product owner as needed. These conversations happen as often as needed and include whoever is necessary. Written documentation may continue to exist when a story-based requirements approach is used. However, the focus is shifted dramatically from written to verbal communication.
An Agile Team Inspects and Adapts
The plan created at the start of any project is not a guarantee of what will occur. In fact, it is only a point-in-time guess. Many things will conspire to invalidate the plan—project personnel may come or go, technologies will work better or worse than expected, users will change their minds, competitors may force us to respond differently or more rapidly, and so on. Agile teams view every such change as presenting both the opportunity and need to update the plan to better reflect the reality of the current situation.
At the start of each new iteration, an agile team incorporates all new knowledge gained in the preceding iteration and adapts accordingly. If a team has learned something that is likely to affect the accuracy or value of the plan, they adjust the plan. The accuracy of the plan may be affected by the team’s discovering they have over- or underestimated their rate of progress. Or they may discover that a certain type of work is more time consuming than previously thought.
The value of the plan may be altered by knowledge the product owner has gained about the desires of likely users. Perhaps, based on feedback from seeing the software from an earlier iteration, the product owner has learned that users would like to see more of one type of feature and that they don’t value another feature as much as was previously thought. The value of the plan could be increased in this case by moving more of the desired features into the release at the expense of some of the lesser-valued features.
None of this is to say that agile teams take an ad hoc view of changing priorities. Priorities do tend to be relatively stable from one iteration to the next. However, the opportunity to alter priorities between iterations is a powerful contributor to the ability to maximize the return on the project investment.