Iterating faster with SAFe

Here’s a frequently asked question in the SAFe community: I wanted to understand what SAFe says about someone who wants to go faster than 2 weeks of iteration? I mean the whole PI concept is based on 5 iterations worth of planning. What if a team/organization wants to develop and synchronize faster than 2 weeks? Is speed going to be compromised by following the standards of PI cadence?

Here’s my take:

Adjusting Cadence Length in SAFe — Can you? Should you?

SAFe considers the 2-week iteration length as a default rather than a rule.

The question you need to consider is what inspection and adaptation cycle you’re looking to accelerate — the Iteration or the PI.

Basically, do you want an opportunity to tactically adjust priorities more frequently than every 2w? Or do you want to adjust a more strategic direction more often than every 8–12w?

With the answer to that, you can experiment either with iteration length and/or PI length. Of course, the cadence length affects coordination overhead — there’s a fine balance.

Additionally, we’re talking about a Planning, Inspection, and Adaptation cadence — NOT the release cadence. Releases are on-demand meaning can be more frequent (or less).

Iteration Goals and PI Objectives provide us with room to maneuver

Another point to remember is that you can adjust iteration backlogs as long as you’re focusing on iteration goals. And even PI objectives can be adjusted — “Assume Variability Preserve Options”. If it’s occasional adjustment it’s not a reason to necessarily use a faster cadence.

Is team-level Kanban the solution to the need for more flexibility in SAFe?

Many teams think Kanban might be the best choice for them if they need more and more flexibility. Kanban CAN be a better fit if your demand is extremely volatile. I would be very careful though. Doing some level of goal-setting and prioritization and planning on a cadence is a powerful way for a team to focus. Do we really WANT to be a strictly reactive team?

Kanban combined with flexibility with some of the capacity we have each iteration can definitely be helpful and is why we recommend all Agile Teams in SAFe use Kanban to limit their WIP and improve their flow — this actually enables them to change scope even within an Iteration if that’s needed in order to achieve their Goal. (see my recent blog post that talks about dynamic scope in SAFe).

“Kanban Teams” in SAFe have an iteration cadence with the establishment of iteration goals even if not detailed iteration backlogs. Maybe that’s a good fit for your context maybe not. It might be an interesting experiment to try.

What if Planning a PI doesn’t make sense?

Finally, if PI planning doesn’t make sense even if PI is shorter — maybe you need to reflect on SAFes appropriateness for your context or on what’s so volatile about the demand coming your way and whether it’s “nature of the beast” or a systemic impediment to work on …

What’s the Bottom Line

What I tried to show here is that a conversation about what to do when the iteration feels too long should start with “Why”. Get to the bottom of what’s currently not ideal, look at the different options, consider Lean, Agile, and SAFe principles, and figure out whether it makes sense to change the cadence, change your approach to the balance between planning and flexibility, the difference between committing to goals and committing to backlogs, and the role that more flow-oriented techniques such as Kanban can play in addressing your issue.