8. 1. Visualize the Workflow
• Map your current process as-is.
• Columns represent process steps for value-added work.
• Include every step from idea to end consumer.
9. Pro-tip:
The board is not permanent. Iterate the board configuration
periodically, as a team, to find the optimal configuration for your
process. Use retros for this.
10. 2. Apply WIP Limits
• Only allow a limited amount of work to be in each column.
• Work does not move into the next column until a space opens up.
• The effect is a “pull system” where the end consumer gradually
pulls work through like a vacuum.
11. Pro-tip:
Despite overwhelming scientific evidence to the contrary, people still
think if they multitask they will get more done, more quickly. It is
simply not true.
12. 3. Make Policies Explicit
• Some work items have higher “cost-of-delay”. Allow them to take
the “carpool” lane, using explicit service classes.
• Most items will be a “Standard” (FIFO) class of service. Others
might be “Expedite” or “Fixed Date”.
• Separate classes of service will smooth flow overall. In business,
predictability can be worth more than speed.
13. Pro-tip:
Reserve some capacity for unexpected work, perhaps with an 80/20
distribution. This will allow for more urgent work to move faster
through the system, without affecting the flow of standard items.
14. 4. Measure and Mange Cycle Time
• The metric for speed is Cycle Time or Throughput. They are
mathematically equivalent.
• Cycle time = the average work time from idea to consumption.
• Throughput = the number of items completed per time period.
15. Pro-tip:
A Cumulative Flow Diagram (CFD)
is the best visualization for this,
but it takes some getting used to.
You can also use a Control Chart,
but it is best to have both.
16. 5. Optimize with Science
!
• Think carefully before you modify the board.
• Make changes to the board based on a stated expected outcome.
• Measure whether it actually improves performance, and change it
back if it doesn’t work.
17. Pro-tip:
Consistent violations of WIP limits at the same column indicate a
communication or coordination breakdown. Half-done work is probably
being stalled or blocked, prompting the premature pull of new work.
19. WIP Limits Increase Speed
• Knowledge workflows, like networks, are subject to the same
constraints as any flow-based systems.
• In systems with variable arrival and processing rates, queues
tend to form between process steps.
• Queue size increases exponentially with linear increases in
capacity.
20. WIP Limits Increase Speed
• As capacity increases, systems risk catastrophic collapse (think
traffic jams).
• WIP limits place an upper bound on the capacity of the system to
prevent exponential growth of queues.
22. Velocity
• Velocity obscures the most important source of waste: queues.
• Most of the waste in knowledge workflows comes from items
waiting in queues between processing steps.
• The relative difficulty of engineering tasks has an insignificant
impact on the overall throughput of the system.
23. Estimates
• Planning poker encourages “group think”. Statistics beats human
intuition every time.
• Uncalibrated gut-feeling estimates are worse than useless. They
are dangerous.
• You can predict with 95% confidence that an item will be
completed within two standard deviations of the mean cycle time.