Thursday, May 15, 2008

Quick Poll Results: Sprint Duration

Following the discussions both elsewhere and on this blog about sprint duration, the questions poses itself - what is the preferred sprint length? So I ran a poll.

And the winner is... 2 weeks.

According to this very scientific survey (1 week on my blog, 228 Visitors, 50 votes), the most popular sprint duration is 2 weeks, followed by (early leader) 3 weeks, which lost by one vote.

The final results are:
1 week2 (4%)
2 week18 (36%)
3 week17 (34%)
4 week5 (10%)
1 month5 (10%)
longer, but fixed5 (10%)
variable1 (2%)

How long should the sprint be? There are many right answers, some of which argued for longer sprints:
  • The longest time you can shield your team from changes in work
  • bigger/harder stories imply longer sprints

But most argued for shorter sprints:
  • It should be short compared to the length of the project
  • Quicker response to changes/new information
  • More and earlier data points for measuring velocity
  • Achieving closure - there is a feeling of success associated with a successful sprint - the more often the better
  • Agility (and ROI) - the shorter sprint sprint, the quicker you can get functionality to the customer or user
  • Feedback - the sprint end is a chance to learn from the customer and from yourselves - the more often the better
  • Reliability of Commitment - the deadlines are nearer, the committments fresher in your memory
Very short sprints, i.e. 1 week are appropriate for very short projects, say 1 month. To do a 4 week project in 1 sprint, or even 2, does not leave much opportunity for feedback.

Mike Cohn pointed out:
3-week sprints have become very popular over the last 12-18 months. Before then most teams considered them odd ;) Seriously, it's a recent change for most teams
And the closing word:
The preferred sprint duration is the one that works best for your team :)
--Dmitry Beransky
I'll second that!

Thanks to Dmitry, Ben, Ash, Ilja, Paul, Kiran, Paddy, and Brett for their comments, as well as to everyone who voted!


savaged said...

I note some teams have a variable sprint duration. Is this a "Scrum smell" or is there any merit in it?

Peter said...

Hi Savaged,

It's considered to be a smell. According to Jeff Sutherland's latest Scrum-But test, an iteration length of 4 weeks or less is good. If the sprint length is longer or variable, this has been found to be a sign of lower performance.