Learning objective
By the end of this lesson, you can distinguish an end-to-end Lean improvement from a local efficiency claim and design balancing measures that expose displaced work or harm.
Why it matters
A team can become faster while the customer waits longer. High utilisation produces queues, automated defects arrive downstream sooner, and a “saving” may appear only because another team absorbs the work.
Core concept: the value stream is the unit of improvement
Judge a change across four dimensions:
- Customer outcome: did the needed result improve?
- Flow: did end-to-end lead time, waiting or work in progress improve?
- Quality: did first-pass quality improve without hidden rework?
- Human and system load: did overload, risk or effort move elsewhere?
A local activity measure is useful only when connected to these system outcomes.
Visual explanation
Text alternative: every local change is assessed against customer outcome, end-to-end flow, quality and rework, and workload and risk before it is called a system improvement.
Worked example
An AI classifier reduces intake handling from eight to two minutes. However, urgent-case recall is untested and downstream exceptions rise from 6% to 14%. The correct conclusion is not “75% productivity gain.” The pilot may have reduced one touch-time measure while worsening quality and downstream load.
The next cycle tests a narrower class with verified labels, an urgent-case control and measures for total lead time, exception rate and reviewer workload.
Common mistake
Mistake: converting released time directly into a cash saving.
Time is a capacity effect, not realised value by itself. State what the capacity enables and what must change for the organisation or customer to receive the benefit.
Quick check
Which result is strongest evidence of Lean improvement?
A. One team processes 30% more cases.
B. Software utilisation reaches 99%.
C. End-to-end lead time falls with stable quality and no downstream overload.
D. A workshop identifies seven wastes.
Answer: C. The whole value stream and balancing conditions improved.
Practical prompt
Choose one proposed automation. Write its local metric, customer outcome, end-to-end measure, quality measure and workload balancing measure. Identify one way the apparent benefit could be displaced.
Summary
- Improve the value stream, not one utilisation number.
- Separate touch time from end-to-end lead time.
- Make quality, workload and risk visible.
- Treat released capacity as a hypothesis until value is realised.
Next lesson
Continue with Make Abnormal Conditions Visible, where 5S turns a workplace standard into operational feedback.