Knowing whether it is working.
WHAT TO MEASURE
Defects or errors, by type Rework, as effort Complaints On-time delivery Repeat problems
WHY BY TYPE
The total tells you little; the pattern tells you what to fix.
WHAT TO ESTABLISH
A definition of a defect that everyone applies the same way.
WHY
Inconsistent counting makes the data meaningless.
WHAT TO MEASURE BEFORE IMPROVING ANYTHING
The current position.
WHY
Without a baseline, improvement cannot be demonstrated.
WHAT TO BE CAREFUL WITH
Measuring people rather than processes.
WHAT HAPPENS THEN
Errors stop being reported.
WHAT TO MEASURE INSTEAD
The process outcome.
WHAT LEADING INDICATORS ARE
Measures that predict problems before they occur.
WHAT EXAMPLES LOOK LIKE
Near misses Steps skipped Backlog growing Overtime rising
WHY THEY MATTER
They allow action before the failure.
WHAT TO DISPLAY
A small number of measures, visibly.
WHY A SMALL NUMBER
Many measures produce no attention to any.
WHAT TO REVIEW
Trends, not individual points.
WHY
Single measurements vary naturally, and reacting to each produces chaos.
WHAT TO ASK OF ANY MEASURE
What decision does this change.
WHAT TO STOP MEASURING
Anything that changes no decision.
