How Big is a Story?
In conversations about Kanban, @dougalcorn mentioned Jeff Patton’s excellent article Kanban Over Simplified. What was specifically interesting about this article to me was The mystery of the shrinking story.
You might be surprised that the term “story”, as described by Kent Beck in the first edition of Extreme Programming Explained, was characterized as “estimable between one to five ideal programming WEEKS.” This was shocking to me, as it would be to most developers. Not so long ago, here in the office, we would say that if a story was on it’s second day, it needed some help, and if it was on it’s third day, there was something wrong! I imagine that most dev shops are similar.
What caused this amazing shift in story size? There are a few reasons. Customers, in order to plan and assess risk, want estimates, and smaller stories are easier to estimate. It’s also natural for developers to fall into the trap of seeing the backlog as a todo list of tasks that need to be completed. I also think one of the main reasons stories have become so small is to maintain the illusion of momentum. It’s about developer happiness, and the satisfaction that comes from clicking “Start, Finish, Deliver” on a daily basis. I’ll talk about why that’s an illusion later, but I personally feel that maintaining that illusion is beneficial and maybe even crucial to a successful project.
What is wrong with small stories? They make the backlog longer, and small stories are harder to reason about and fit together into anything cohesive. But the real detrimental effect of small stories is loss of focus on the customer’s needs. Small stories lean more towards up-front planning and implementation, rather than customer goals and values. The illusion of momentum is that forward progress is being made, when no value is being delivered. It is not uncommon to break a feature down to such a detail of implementation, that the product owner is asked to click “Accept”, having no confidence that anything was actually done. Asking a customer to “take our word for it” is fail.
Don’t get me wrong, planners should still strive to make stories as small as possible. However, stories must first and foremost meet a standard of minimum marketability. A minimally marketable feature is “the smallest possible set of functionality that, by itself, has value in the marketplace”. Meeting that standard is not always easy to do in a few days of development.
Most importantly, stories that the customer sees should clearly define value. The most effective way of doing this is to write a story using the format described in Cohn’s User Stories Applied “As a [type of user] I want to [perform some task] so that I can [achieve some goal]”.
But what about developer happiness and momentum? Isn’t it demoralizing to be working on delivering one piece of functionality week in and week out? Isn’t it satisfying to click “Start, Finish, Deliver” on a daily basis? I would argue that it’s equally as important for developers, behind the scenes, to continue striving to deliver subsets of functionality on a daily basis, and to track that. I think we have a lot to learn about how to achieve that and also focus on customer value.
In our previous article, Why Not Pivotal Tracker, Doug states correctly that velocity is flawed. I still think that the illusion of velocity is a beneficial force from a developer standpoint. Velocity is a lie that deceives a customer, but motivates a developer. Perhaps we need new terminology to deal with that separation of concerns. One thing is clear, first and foremost, “stories” should define and communicate value to the customer.