Guideline: Use Case Formats

Use Case Formats

Use cases differ from project to project and person to person. A use case that works in one situation may be totally unsuited for another. Different projects have different needs. (See [ADO04] for more information on use case formats.)

Some need rigorous documentation, including high-ceremony use cases, which are formal, highly structured use cases. If the writers used a template, then they filled out all or almost all of its fields for each use case. High-ceremony use cases are best suited for large, extremely complex, safety-critical systems, such as flight control systems, telephone switches, and so forth. They are also used in development cultures that have high documentation standards.

Other projects may be more agile and less formal, benefiting from low-ceremony use cases, which are informal, less rigidly structured use cases. If the writers used a template, then they may have left many of the fields blank. Low-ceremony use cases are best suited for smaller, less complex, less safety-critical systems where most of the stakeholders have a strong background in the problem domain. Sometimes, simple descriptions suffice, such as use case briefs.

As Guideline: Detail Use Cases and Scenarios explains, it makes sense to write use cases iteratively. Starting with the basic details, you can then identify the various alternative and error paths that the use case might follow so that you can evaluate, rearrange, or eliminate them, and then elaborate or fill in the details of the courses that you intended to use. You can then write the use cases in one or more of the following formats, progressively, until you reach the one with the level of detail required for a specific project:

Actor-Goal list: A format for the overview
Briefs: A format for writing summary use cases
Improvisational score: A format for writing less formal, low-ceremony use cases
Symphonic score: A format for writing more formal, high-ceremony use cases

Actor-Goal list

Context: You have identified your actors and are trying to identify use cases.

Problem: Developing a set of use cases in an ad hoc manner can lead to unnecessary work, missing features, and feature creep. Weight is one of the most important factors in space flight — so important that the United States space agency, NASA, will not allow anything on a spacecraft that isn’t absolutely critical to the flight. If something literally isn’t worth its weight, then it doesn’t go. Likewise, each use case adds cost to a system; therefore, you need to be sure to include only those use cases that add some kind of value to your collection.

Forces:

Simply listing actors or listing goals is not informative enough, but actors and goals together are informative. The classical approach to writing use cases is to define a list of actors, then find use cases for each. A variation on this theme is to itemize what the system must accomplish. Yet, neither approach is adequate by itself. You need to know both who is using the system and why they are using it. Otherwise, you introduce the potential of either feature creep or missed features. At the least, a set of use cases should describe this association.
A quick overview of the entire project structure is sufficient and necessary early in the use case development cycle. Ideally, this overview should be as short as reasonably possible. It must contain key information as to who requires each service and why they need it. Most other information is not very useful at this stage of the project, because it runs the risk of quickly becoming obsolete, as well as discouraging out-of-the-box (innovative) thinking. An overview helps the writers work through the entire set from a high-level view, expanding some use cases, eliminating others, and combining still others into a more logical grouping.
You need to be able to expand each to a full use case on demand. A seedling use case forms the basis for a full use case later in the iterative development cycle. Each seedling use case needs to convey enough information so that someone, possibly other than the outline writer, can easily go back and expand it into a more informative use case.

Solution: Build an Actor-Goal list, which is a list of actors and their goals that gives you an overview of entire project needs.

Start by identifying the list of actors who will use the system, and then identify at least one goal for each. Actors without goals indicate that you haven’t adequately defined the system. The actor is beyond the system’s scope, doesn’t belong in the system, or is part of another actor.
Likewise, leftover goals can indicate that the system is too complex and you're trying to accomplish too much, or that you haven’t adequately defined all of the necessary actors. Carefully evaluate the leftovers to see if you are just overlooking some detail, or whether they don’t belong in the system.
Remove unassociated actors and goals from the list.

Sometimes, this list may provide enough information to serve as use cases for very small, high-communicating, low-ceremony project teams. Usually, the actor goal list is the first step of identifying use cases.

Briefs

Context: You have written an Actor-Goal list that outlines your use cases.

Problem: Relying solely on an overview to capture the important parts of a system’s behavior is dangerous, because it provides only high-level information and can easily introduce ambiguity into a system.

Forces:

Although valuable, an Actor-Goal list does not clearly describe a system. Usually, an outline doesn’t provide enough precision to avoid ambiguity, which can wreak havoc on a project by leading to unnecessary or erroneous development. Yet, an outline is helpful, because you still want an overview that you can easily scan. Unfortunately, with the passing of time or sheer volume of work, it’s too easy to forget details that were obvious to you earlier.
Iterative use case development requires creating placeholders for expansion. To develop use cases iteratively, you start with sparse use cases, reorganize them, and flesh them out as the system takes shape. Ideally, these placeholders should be clear enough to: 1) unambiguously describe their role in the system, and 2) allow someone to expand the use case, even if they were not involved in writing them originally.
Because outlines are general by nature, do not spend a lot of time, energy, or money writing them. Outlines provide an inexpensive method of documenting complex ideas in a manner that is easy to follow, and they provide a mechanism for people outside of a project to understand the high-level concepts. While it may take some effort to think things through, you don’t want to waste resources describing your ideas. The system is still in a state of flux at this point, and it is too early to spend much time documenting its shifting details.

Solution: Write two to four sentences per use case, capturing key activities and key-extension handling.

Expand the Actor-Goal list into briefs by writing a two- to four-sentence use cases for each entry in the list.
Briefly describe each use case’s main scenario and most important extensions.
Include enough information to eliminate ambiguity for at least the main scenario of the system.

Improvisational score

Context: You are operating in well-known domains or in situations where writing high-ceremony use cases would require all of your allotted development time.

Problem: Writing formal, high-ceremony use cases when lesser detail would suffice wastes time and resources.

Jazz is considered to be “musician’s music,” and jazz players are usually highly skilled. Many jazz musicians prefer to improvise in small, highly skilled teams, such as jazz quartets. To improvise effectively, the musicians must have a thorough understanding of the conventions that form the given musical style, including chord sequences, rhythmic patterns, and melodies. These conventions provide a basic framework for the musicians to interact as a team, while still allowing room for spontaneous creativity.

Likewise, use cases do not always need to be specified in excruciating detail. A far-preferable strategy is simply to define the basic structure of what the developers need to implement. The use cases act as placeholders that may be elaborated later or simply improvised by the developer who implements the use case.

Forces:

Briefs do not provide enough information. While useful, use-case briefs describe only the more significant parts of behavior. Often, developers need more information, especially when working in unfamiliar domains or in the heart of the system, where the actor has many choices to make and many paths to follow. Briefs do not describe all of the important events that can happen, nor do they describe the details that go into making choices along the way.
Fully elaborated use cases can be too expensive, time consuming, long to write, and boring to read. It takes a lot of time and effort to write a formal, fully descriptive set of use cases. Maintaining this set takes even longer. Often, a collection of use cases reaches the point of diminishing returns long before it is completely written, much less formalized. Readers often prefer shorter, simpler use cases over long, complicated ones, because overly detailed use cases can be overwhelming and, frankly speaking, quite boring.
Many groups communicate well enough to resolve ambiguities on the fly. While briefs may be insufficient, stakeholders don’t always need everything to be spelled out for them. Developers are usually capable of asking questions and filling out the necessary detail from their own domain knowledge. Many people can work with a fair level of ambiguity, and most organizations possess what is often referred to as their “core competencies.” Mature organizations with strong domain knowledge can survive, and even thrive, using more informal, less precise use cases.

Solution: Specify the use cases at a low level of precision, allowing the developers to fill in the missing details as necessary. The level of precision required depends on the background experiences of the development team. Skip the less meaningful fields on the template, and write the Main Scenario section as a simple paragraph. Describe key-extension handling in the next paragraph or two. Be prepared to resolve ambiguities and expand detail on the fly throughout the project.

When you can rely upon open and frequent communication among the developers and customer, write the use case with less detail and precision. The developers can fill in the gaps by asking users or by using knowledge of the domain. However, the developers need a thorough understanding of the business context to be able to fill out the details themselves. Even the most knowledgeable developer will still need access to the customers and users to get answers to questions and clarify requirements.

Ideally, the project will be structured to enable effective communication between the customer and the developers. Typically, this will involve having a small, co-located team, with the developers having easy access to the users throughout the project. The risk of misunderstanding can be resolved by frequent incremental delivery if the development organization has a relatively low-ceremony culture.

Jazz improvisation does not always work. It can become tedious and unpleasant to listen to, even for the committed connoisseur. For this reason, you also need feedback from the audience to determine the success of the improvisations. Multi-level or two-tier reviews are critical to success (see Guideline: Effective Requirement Reviews).

Improvisation may not always be suitable for the organizational culture, a full symphonic score may be preferable in large, high-ceremony teams (see section that follows). For instance, I once watched a conductor toss his baton away in disgust when a pianist improvised to such an extent that the orchestra could not follow the score. If the organization deems the risk of improvising to be unacceptably high, then you can specify the use cases with a higher level of detail and precision. You could start with a strategy of specifying low levels of detail and precision, and then adapt as necessary.

Symphonic score

Context: Writing structure for high-ceremony situations, such as when there are many developers or when development teams are geographically dispersed.

Problem: Writing low-ceremony use cases for high-ceremony situations raises the risk of miscommunication to unacceptable levels.

A conductor’s version of a symphonic score contains the music for the entire orchestra, as well as any accompanying vocals. The parts to be performed by different voices or instruments are written on a separate staff, with all of the staves aligned, one above another. This score specifies each note and its associated timing in precise detail, so that the orchestra can perform a symphony as the composer intended.

As with use cases, a score tells the musician what to play, not how to play it. For most symphonies, the orchestra will not be able to meet the composer, so instead, they must rely upon the director to interpret the score and the composer's intentions.

Forces:

Certain development situations and cultures require high degrees of formality. Some organizations operate in a highly formal manner, thus require a highly formal process. While this formality may not be desirable, it is the company's way of doing business, so things need to be done that way. Other organizations are highly formal because they do highly complex, life-critical work, where even small failures could have disastrous consequences. For instance, no one would feel comfortable flying on an airliner with an off-the-shelf, one-size-fits-all flight management system.
The cost of repairing miscommunication is high. It is easy to write vague, inadequate use cases full of ambiguity. Use cases can be too brief and ambiguous, or contain domain-specific details that may be beyond the understanding of many stakeholders. Either way, they provide an opportunity for a misunderstanding that leads to an incorrect implementation. The cost of correcting these mistakes depends on when they are discovered. Earlier is cheaper than later, especially when later means customers finding the problem in the delivered product. To avoid miscommunication, aim to write use cases that are general enough for all of the stakeholders to follow, yet precise enough for the developers to use when building the system.
Developers need detail for implementing steps, business rules, data fields, and, especially, for handling extensions. No one has developed a program that can take a set of use cases as input, and churn out a completed system. Even the best-case tools seem to require human intervention to flesh out details and resolve ambiguities. Similarly, developers who do not understand the business context or lack domain expertise may not be able to fully comprehend a product. In an ideal project, software developers would have access to the domain experts to ask questions, so they could fill in any areas that may have been missed (see Improvisational score, previously). But often, they do not ask. Therefore, they misunderstand the more complex or ambiguous use cases in the set. To develop a system correctly, a team needs either access to domain experts or additional information that describe the steps, business rules, data fields, and extension handling that they are implementing.

Solution: Specify your use cases with a high level of precision, explicitly filling in all of the details in the use case template, while staying technology-neutral. The level of precision required depends on the background experiences of the development team.

Intuition may tell you that if some detail is good, then more must be better. However, be careful about falling into the trap of over-specifying details. It’s naive to believe that everyone who reads your use cases will be able to understand them. Different people may interpret the use cases differently. Prepare for this eventuality in your process, and avoid the tendency to over-specify your use cases. If you try to specify a use case in too much detail, you may fall into the classic analysis paralysis trap.

People are often tempted to address the communication problem by trying to explain the business domain within the use cases. In a similar manner, they include too much technical detail. Succumbing to these temptations by explaining the business domain or including technical details is always a mistake, because it complicates the process and obfuscates the requirements. The reader of the use cases cannot distinguish the real requirements from the boring background information, so will soon get distracted and lose interest. Instead, include this information in an extra section.

If you are handing over the requirements to a development team whose members are unfamiliar with the domain, then you will need an alternative strategy for teaching them the domain knowledge.