WIP drives Lead Time and Flow - 130519
Transcript of WIP drives Lead Time and Flow - 130519
WIP drives Lead Time, Delivery Rate, Flow Efficiency and Quality
Amit Kaulagekar
´ Based in Pune, India
´ [email protected]´ Linkedin: Amit Kaulagekar´ Areas of expertise –
´ Lean, Kanban and Six Sigma´ Project and Program Management´ Service Management´ Business Process Management´ Standards and Compliance´ Software Engineering
Let’s start with…
´ Why do we want to adopt Kanban?
´ Achieve shorter lead time
´ Higher delivery rate
´ Better flow efficiency
´ Better quality and
´ Better Throughput
´ OR....
´ All of this
Kanban Metrics
´ Lead Time (System Lead Time / Customer Lead Time) – observe time it takes to fulfill requests from the point of commitment
´ Throughput (completed items per unit of time) – measure completion rate
´ Flow efficiency (work time over lead time)
´ Failure demand (self-demand caused by a failure to do something or do something right for the customer)
´ Initial quality (escaping defects that will generate future failure demand)
´ Blockers and their impact
´ WIP levels (number of items in the system overall or in certain stages)
´ Discard rates (items that were discarded before and after the commitment point)
Productivity in Kanban
´ Count the number of tickets completed in a time period, and pair that to the flow efficiency. The counter-measure to this is to measure local cycle team and quality (or tickets which move backwards due to quality problems)
´ If you optimize for minimal local cycle time (in an activity) coupled to high quality (minimal defects passing down the line, or minimal tickets returning) at the local function level, and use flow efficiency at the workflow/service delivery level then together they are very powerful to drive system-wide improvements.
´ Delivery rate metrics should not come at the expense of flow efficiencyand hence these can also be treated as a vector - you can boost throughput by swamping a system with WIP or by sacrificing quality so you need counter-measures to discourage this. Measuring overall lead timedistribution is also a good counter-measure as well as generally being a customer-valued fitness criteria metric.
WIP
´ When WIP is Higher´ Higher / longer Lead Time´ Higher Delivery Rate (initially and then it reduces gradually. No linear relationship though)
´ Lower quality
´ Lesser predictability
´ And controlled WIP results in ´ Lesser lead time
´ Better predictability´ Better quality but
´ Lower delivery rate (some call it productivity)
Impact of economic drivers
´ If your WIP is high, reduce it
´ First minimally and then gradually
´ If your WIP is already low, consider your economic drivers
´ If Productivity drives your bottom line, don’t push WIP too low
´ If time to market or quality drives your bottom line, push WIP as low as it will go
So what is the goal…
Let’s test this in a controlled environmentWith GetKanban game
Little’s law
Avg. Lead Time
Avg. Delivery Rate
WIP
Poolof
ideas
Readyto
deliver
Delivery Rate(from the kanban system) System Lead Time
WIP=
Let’s calculate
Team started with WIP limit of 12, increased it to 16 and again brought it down to 12.
Delivery rate on• Day 10
• (WIP / Lead Time) = Delivery Rate• 10 / 7.33 = 1.36
• Day 12• 13 / 9 = 1.44
• Day 14• 20 / ?????
Run chart and Lead Time Distribution chart
We use Lead Time distribution to calculate average lead time for ‘Delivery Rate’ calculation. In the given scenario, Avg. Lead time is 8.75 days
And Run chart helps us find trends / patterns in the process. As we can see, variation is very high in this process.
Flow Efficiency and Finance
Out of 19 delivered tickets, team could calculate ‘Flow Efficiency’ for only 1 ticket which is 50% (during12 day time period).
And the economic driver was shorter lead time or ‘Time to market’
So here are the numbers for all days… and trend
‘Time to Market’ being an economic driver, by increasing the WIP, team achieved exactly opposite of it.
Another example where WIP was 20
• Work is moving in batches• Lead time distribution is skewed towards right• Run chart shows high variation• 45% Flow efficiency but• Delivery rate is on a higher side
Flow efficiency
When WIP is 10
When WIP is 12
With capacity allocation (Intangible)
WIP limit can be´ by person,
´ By workflow,
´ By work item type,
´ By total number of items in progress
´ By risk dimension or
´ Across the entire board
´ Aggregated personal Kanban
´ Team Kanban
´ Aggregated team Kanban
´ System Kanban
´ Discovery and Delivery Kanban
´ Service Delivery Workflow
For
And how do we decide WIP
´ Team size + some buffer (like 50%)
´ Twice / Thrice the work items per individual
´ Xxxxxxx
´ Most of the times, it is just a number at the top of the column
And what do we want to achieve?
´ Better / Shorter lead time
´ Better predictability
´ Better quality and
´ Less delivery rate
´ Better delivery rate
´ Longer lead time
´ Less predictability and
´ Less quality
OR
Economic drivers
System Kanban
System Kanban in the context of end-to-
end flow
Sourse: Essential Upstream Kanban’ by Patrick Steyaert
Implementation
´ Agree on economic drivers
´ Find optimum level of WIP
´ Stop analyzing metrics in isolation
´ 3 to 5 months to see the change (team size)
´ Linking it with ‘Class of Service’ and ‘Cost of Delay’
´ Arrival rate of new work
´ Size of work items
Acknowledgement
´ David J Anderson
´ Frank Vega
´ Janice
Thank you