Optimal Length of Agile Sprints

Agile is a popular development methodology sometimes it seems to be the only method in use today.

At the core of Agile is to break work down into small chunks that can be managed by the scrum team in a short period of time known as the sprint.  Ideally the sprint delivers some or several useful functions that can be shown (demo) to stakeholders and ideally to customers in order to gain feedback and to change and improve as quickly as possible. This objective is often to deliver MVP or Minimal Viable Product although this is in our experience often beyond the realistic scope of a single sprint.

The idea is that the team commits to a given quanta of work that can be experienced by the customer and get quick feedback. It also allows flexibility to change direction or priorities of the overall development.

In our organisation one of the big questions is the ideal length of the sprints.  

To set the context most of our development does not get released to customers on a one or two sprint basis. We use sprints as a meaningful checkpoint and as a way of involving other teams, stakeholders, sponsors and management.

Previously we used three week sprints and in this specific business unit the custom has been sprints of two weeks.  We are now in the process of moving to three weeks; partially at my initiative.

The reasoning is that two week sprints are valuable when the output does get delivered to users and especially if it is user interface rich. Fast work and rapid feedback and intense market dynamics seem to fit in this context.

However for work that develops over time especially with a lot of infrastructure or core development then there are many disadvantages to short sprints.

Agile sprints have high overheads in each and every stage of  preparation, execution and post sprint analysis (Discovery, Pre-planning, Planning, Retrospectives etc.) There is also management overhead in driving an organisation with several projects each with short sprints and low capacity. Overall these overheads that amount to high costs that the organisation can not afford over time. Longer sprints have lower overhead and so are more efficient. Our analysis indicates that moving from two to three work sprints cuts most of the overhead linearly.

Shorter sprints present a challenge in trying to deliver real value (MVP or MVP-like) even more so if the team is quite small. It is perceived that the small capacity presents its own challenges in preparing and planning the sprint trying to customise the business needs to the limited time.

Another critical issue is the vulnerability to planned and unplanned absences. It only takes a couple of days vacation, course or sickness to wipe out the content of a sprint. Bugs, tricky integrations and environment issues have a similar disastrous impact.

Finally, short sprint cycles place a large demand on the Product Manager and his team. The cycle times are so short that there is never any time to pause and consider the bigger picture and to research new ideas and future directions. Again we think that there are almost linear savings available by increasing the sprint length.

Overall short sprints are considered expensive, hard to plan and prone to delay. Our analysis indicates that three  week sprints are a better balance of flexibility (Agility) and efficiency.