During the last year or so, I have been caught up several times in discussion regarding ”agile requirements” or some similar catchy phrase. ”Christoffer”, they say, ”as an agile coach, what is your take on requirements? Should we have them? In what format? Who should write them? Where should we store them? And can we really start developing stuff without having them ready and approved?”
The problem with these questions is that there really is no straight, ”do-it-like-this-and-all-will-be-well” answer.
I’m writing this post in a hope to address these questions, and more.
Requirements are neither the paper nor the ink
Let’s start by clarifying an assumption, that is often not fully understood. A requirement is not a paragraph on a piece of paper in a binder on a dusty old book shelf. Neither is it a user story, entered onto the back of an index card on a Kanban board. Those are not requirements – those are the artifacts of the requirement.
A requirement represents knowledge of a need or want, presumably related – at least on some level – to a system that an agile team is (or will be) working on. You need that knowledge in order to develop something. You cannot develop something without requirements.
How do you extract this knowledge? That depends. Usually talking to the person(s) with the need/want helps. Workshops or surveys might also be viable tools.
Who extracts this knowledge? Someone with the required skill set, obviously – but you already knew that. Does that mean that you need a requirement analyst in your team? No, not necessarily. Ask the team if they are comfortable with doing the requirement analysis themselves. Perhaps they have both the technical skill and the communication skill required.
How do you document this knowledge? Again, that depends – it could be in the form of use cases, user stories, mock-ups, prototypes, or whatever suits that particular form of knowledge. Document only when there is a clear purpose with the resulting artifact. If everyone involves have a sufficient understanding of the requirements, there is typically no need to document anything.
The artifacts are still important
Here is a huge, potential pitfall. Because, typically, we believe that we have sufficient understanding of the requirements, while we actually do not. We believe that we have collected all possible knowledge of the need, while we in fact have not. Or, it could be that the need changed, as they often do.
One purpose of creating requirement artifacts is thus to create a common ground, as a help to capture this knowledge and use it as a basis for discussion. But be aware that the use case or user story or whatever you use will change many times – because that is, after all, one of the purposes.
And, obviously, agile methods to the rescue: create something small, based on a few of the requirements – as far as you understand them at this point in time – and then show the result to the one with the need. This will help everybody’s understanding of the actual need, and lets you adjust the end result as you go along.
A philosophical view
A requirement can be viewed as a part of a perfect – albeit constantly changing – knowledge about a need. The artifact used to capture this knowledge (be it that dusty old binder or that index card) will therefore always be fundamentally flawed. There is no way that an artifact, however detailed, can capture all that knowledge. So let’s not try.
Instead, embrace the fact that requirements evolve and change. Thus, your requirements – in whatever way you choose to document them – will not be done until the need is gone. Let’s not wait with development until then.