Archive for May 2015

Velocity—it ain’t what it sounds like

In the physical world, velocity is a common and well-understood term. In the world of Agile software development, “velocity” means something that, at first glance, seems similar to the physical world meaning, but which is in fact different in an important way.

Velocity in the physical world

Anyone who drives a car is intuitively familiar with the meaning of velocity. It means speed, and is commonly expressed in miles or kilometers per hour. A slightly more abstract expression would be some distance per some unit of time. An even more generic way to say that is the ratio of benefit to time (benefit/time).

In the physical world, whether from the point of view of a commuter, a cabbie or a race-car driver, increased velocity means increased value—in more technical terms, that’s an increase in the benefit/cost ratio. In these cases, the benefit is miles traveled. Time taken is the cost.

To the commuter, traveling a fixed number of miles in less time provides increased value by lowering the time required to get to or from work.

To the cabbie, who effectively gets paid by the mile, covering more miles in a shift provides greater value.

To the race-car driver, winning the race has more value than coming in second.

Velocity in Agile software development

People engaged in Agile software development also use the term “velocity” and, probably because of the physical world meaning of the term, can be inclined to believe that increasing velocity means increasing value (or benefit/time), just like it does in the physical world.

Velocity in software development is typically expressed in terms of (story) points per iteration. At first glance this seems very velocity-like—something delivered in some period of time. But a closer look at what is being delivered reveals something interesting.

Story points don’t represent benefit

Many agile teams use relative estimation in one form or another to project how much time it will take to develop various functional bits (user stories) in a product. It’s common to assign a number of points to each story being estimated. In relative estimation, when we estimate a story at 3 points, for example, we’re saying that we believe it will take 3 times the amount of time to develop that story as a story estimated at 1 point.

So let’s say that we work in 1-week iterations, and at the end of a week we total up the estimated points for all the stories we completed that week and find that we completed 25 points worth of stories. We call that total our “velocity” and we use that velocity figure (25 points per week) both to project future completion dates and to determine how much work we think we can do next week.

But what we call “velocity” in software development isn’t like velocity in the physical world, in that points don’t represent a benefit. At best, what we call “velocity” is actually only a ratio of relative time to absolute time. It’s a useful ratio, to be sure, but only so far as it lets us make calculations that convert our estimates to actual (calendar) time. Multiplying “velocity” by the number of weeks remaining until a deadline tells us how many points worth of stories we can expect to complete. Dividing the total number of story points in our backlog by “velocity” yields the number of weeks remaining until anticipated completion.

What we call “velocity” in software development doesn’t represent a benefit/time ratio like it does in the physical world. The largest implication of this is that increasing velocity does not necessarily translate to increasing value.

What’s in a name?

It’s important what we call things. The term “velocity” is certainly more convenient than something like “relative to absolute time factor” but it can mislead if not well understood.

Have you been in a situation where you’ve been encouraged to “increase your velocity”? It’s almost certain that the person doing the encouraging was really asking for you to go faster—to increase the benefits delivered in a certain amount of time.

In hindsight, it would probably have been better to use the term “velocity” to describe a benefit/cost ratio, like it does in the physical world. One of the easiest benefits to understand, at least in a for-profit company, is revenue. Revenue is a benefit, and time is a cost, so revenue/time is a reasonable way to express velocity. Or would be, if the term didn’t already have an accepted meaning.

We may have to live with the term “velocity” to describe what is actually only a “time factor.” After all, we’ve got roughly a decade’s worth of inertia supporting this now-traditional usage.

But we can at least be aware, and strive to make others aware, that increasing velocity isn’t something to aim for. Try increasing value instead.


Clicky Web Analytics