Kanban from the Inside: 5. Flow

This post is one in a series of excerpts from my book, Kanban from the Inside. Chapter 5 is on flow. Like the preceding chapter, it is inspired by the third of the Kanban Method’s core practices, Manage flow.

CP3 (expanded): Manage flow, seeking smoothness, timeliness, and good economic outcomes, anticipating customer needs.


Organizations often get deeper into trouble because people think that the answer is to keep trying harder. They think that better project management will fix a problem of capacity management (scapegoating project managers, meanwhile), that stronger functional management will improve end-to-end performance (when it can easily make it worse), or that people should just try to do better (when the system is fundamentally unreliable).

Kanban’s unusual solution to this problem isn’t to address head-on the roles of project management and functional management; certainly it does not set out to replace them immediately with other things. Instead, it gives managers (and others, too, of course) the tools to see work and how it flows in new ways, together with controls on WIP that impact dramatically on problems of delay and unpredictability. Allowed the right scope, the improvements and the necessary new thinking grow hand-in-hand. If roles then need to change, fine!


It would be a mistake to think that managing flow is only a matter of removing impediments as they arise. Typically in knowledge work, we see work items vary widely in both content and value, and we see the overall workload varying greatly over time. This means that there will always be a place for managing work proactively:

  • Unusual risks and dependencies must be identified early and managed effectively.
  • Looking toward the medium term, anticipated workload must be met with adequate capacity.
  • Longer term, entirely new capabilities might be needed.

Being committed to self-organization doesn’t mean that you must always shy away from managing the most important work items more carefully. Not all work items are alike, and some are considerably more deserving of management attention than others. Dates—meaningful dates, at least—should be met. And when it’s justified by the business opportunity, it’s perfectly valid to sacrifice a little predictability and allow high-value items to jump the queue.