Use Cases Are Not the Only Requirements

Author: Geri Schneider Winters

With all the focus on Use Cases, it can be easy to forget that they are not the only kind of requirements. Use Cases are very good for some kinds of requirements, but really bad for other kinds of requirements. One of our challenges as Business Analysts is to determine the right way to document the requirements for a particular project.

When starting on a new project, one of the first things I want to find out is “Who or what is driving the requirements for this project?”. I was really reminded of this when I read Richard Denney’s book “Succeeding With Use Cases: Working Smart to Deliver Quality”, a book I highly recommend.

If the project is being driven by changes in processes, then Use Cases probably make a lot of sense, since they are good at describing processes. But you could also use User Stories, Scenarios, Flow Charts, or Test Cases to describe processes. Which you choose will depend on corporate culture and the software development process that is used on your project.

In this case, find out what is driving the process change. If the process change is coming from the way that people work, then you can find requirements by talking to the people or watching them work. If the process change is coming from a technology change, then you may need to start working on the requirements by learning about the new technology. I worked on one project which was to design software to run on hardware that tested IC chips. The source of my requirements was the tests that had to be run on the IC chips. I documented those tests as Use Cases because that is what the customer wanted me to use for requirements.

If the project is purely a change in underlying technology, then a lot of your requirements will be the non-functional requirements (often called FURPS). You might have to write these requirements as formal “Shall” statements. Look over a template for a Supplementary Specification for ideas of the kinds of requirements you might need.

For example, if I am installing COTS (commercial off-the-shelf) software, then I do not need Use Cases for how the software works. I might need Use Cases for how we will use the software in our company, if that information is not obvious. But I really need to know other kinds of information, such as how many users will use it at a time, what response is required, how many transactions per second it will have to handle, how many records will it have to store and process, and what kind of security is required to use it. I will have to talk with the folks in Enterprise Architecture to find any restrictions on where or how it can be installed. There may be a need for a new server, so I have to find out the contraints such as what platforms are allowed, or how much money we can spend on it.

In embedded systems, you may find that truth tables or state machines do the best job of describing the requirements. In this case, there is probably no human interaction, and the processes are all closely related to the hardware and how it works. A truth table or state machine is a good way to describe the different states of the system and how it is allowed to change.

Here are some ideas for kinds of things I have seen used (successfully) as requirements for different kinds of projects. Consider these to see if they are appropriate for your projects:

  • Use Cases
  • Shall requirements
  • Test Cases
  • User Stories
  • Scenarios
  • User interface designs
  • FURPS requirements
  • Prototypes
  • Flow charts
  • Activity Diagrams
  • State Machines
  • Truth Tables

 

How do you decide what kinds of things to use for requirements on your projects?

What other kinds of things have you used for requirements for your projects?