The Scrum Guide calls sprints the "heart of Scrum" and also indicates that they should be one month or less. The Guide also cautions about setting sprint durations longer than a calendar month to ensure that inspection and adaption is happening frequently which will reduce the risk of wasted team effort or of building the wrong product. This aligns with the third principle from the Agile Manifesto which encourages teams to deliver value frequently, from a couple of weeks to a couple of months.
A research survey conducted by Scott Ambler+Associates found that the most common duration used by teams following a sprint-based delivery model is two weeks, but how should a team decide what is the optimal sprint length for them?
Their decision should consider a number of factors including:
- How "new" the team is
- Their delivery process
- Constraints on their being able to complete product backlog items
- The maturity or capability of the team to disaggregate work items to a small enough size so that they can complete multiple items in a sprint
- Expectations from the stakeholders who will participate in sprint reviews
For teams which have just formed or are working with technologies or a product which is new to them, too short a sprint duration may prevent them from being able to complete anything meaningful. This could cause them to get discouraged and might frustrate the stakeholders who are expecting to see progress every sprint. On the other hand, if the team is new to flow-based delivery approaches and they start with a long sprint duration, they could revert to traditional behaviors where they complete batches of work items in phases over the life of the sprint.
When in doubt, it is usually better for a team to choose a shorter sprint duration as this should encourage them to identify, elevate and eliminate the real constraints which limit their ability to deliver. With the longer duration, while they might be aware of the constraints, their sense of urgency may be less.
The team should not change sprint lengths on an ad hoc basis, but if they have identified triggers which suggest that they need to increase or decrease duration then such changes might be done after a major product release. Improvements in the team's productivity or a high degree of volatility in the product backlog are two possible triggers. Another example is if the stakeholders attending sprint reviews consistently feel overwhelmed with the content of the product increment being demonstrated to them.
Sprint duration is another case of Goldilocks and the three bears - finding out what's "just right" will take some trial and error!