Everyone loves features. People wouldn’t use your software if it couldn’t do something for them. A lot of the time, the killer feature is the reason the software was created. You have a problem you want to solve, and you have some idea how to solve it. Other times, you’ve already got a product, but you want to make it more powerful or more flexible. It can be difficult not to get carried away when you’ve got an idea for a new feature, but it’s important not to rush into development without proper consideration. I’ve recently been reading Code Complete by Steve McConnell and it contains an extensive list for checking whether your project requirements are missing any obvious holes.

Inspired by this, I put together a heavily modified version of the list to use at work. It’s smaller and more tuned for the work that I do, which is web development. It’s more focused on adding a new feature to an existing project than it is on building a requirements list for a new project.

What Does It Need To Do?

  • How does the user interact with it? What sort of interface might it have?
  • What’s the output? What’s displayed to the user, and in what format?
  • Why do they need this output? Is it in the best format for their needs?
  • What data is needed from the user? What data is needed from other parts of the system?
  • What sort of input validation will be needed?
  • Does the new feature clash with any existing features?
  • Is this similar to an existing feature? Is a new feature more suitable than a modification to an existing one?
  • Does it meet all of the user’s needs? Are all use-cases accounted for? Is it intuitive?

How Good Does It Need To Be?

  • What’s the expected response time? How fast do you expect the feature to be?
  • What’s the acceptable response time? How slow can you get away with it being?
  • What authentication is required? Will the feature behave differently for different access levels?
  • What are the consequences if the feature breaks? Is there any data that must be preserved?
  • How are you going to detect and recover from errors?

Are The Requirements Good?

  • Are the requirements written from the user’s perspective?
  • Do any requirements conflict with each other?
  • Do the requirements avoid dictating the design?
  • Are the requirements sufficiently detailed?
  • Can each requirement be understood by a developer who wasn’t involved in its creation?
  • Is each requirement testable? Can its completeness be confirmed or denied?

Are The Requirements Complete?

  • Is there a record of any unanswered questions?
  • If all of the requirements are fulfilled, will the feature be acceptable to the user?
  • Are any requirements present just to appease someone? If excessive bikeshedding occurred, requirements that aren’t strictly needed could have made it into the list.
  • If all of the requirements are fulfilled, will you be proud of the result?

By making sure I’ve at least attempted to answer each of these questions before I begin typing, the route to a high quality finished product will almost certainly be quicker and more pleasant than if I’d just rushed in headlong.