Stage 4 · Sustain · 08 · 25 min
Measure results against targets
Go-live is not the finish line. The charter is.
The redesigned process has run for a full quarter. This module checks it against the targets the managing director signed in module 5: with KPIs that are defined properly, a scorecard, a look at the automated steps and a control chart that tells real change from noise.
Apply
1. Make sure each KPI is complete
What you do here: separate complete KPI definitions from incomplete ones.
A complete KPI has three parts: what is measured per order, how it is aggregated over a period, and the target. Without all three, two people can read the same number differently.
Which KPI definitions are complete?
Median lead time from enquiry to invoice, per month, target 21 days
5 more to place
Complete
Measure, aggregation and target
Incomplete
Something is missing
Pair every speed KPI with a quality KPI. On-time delivery alone can be gamed by shipping parts that have not been inspected; first-pass yield next to it closes that door.
Apply
2. Compare results with the targets
What you do here: read the scorecard and the realised benefit against the business case.
Custom order process, the quarter after full rollout. For lead time and cost, lower is better.
| KPI | Baseline | Target | Result |
|---|---|---|---|
| Median lead time, days | 28 | 21 | 20.6 |
| Shipped by the promised date, % | 58 | 85 | 86 |
| First-pass yield, % | 85 | 92 | 88 |
| Process cost per order, EUR | 1240 | 1050 | 1070 |
| Customer score, 1 to 5 | 3.4 | 4 | 4.1 |
Realised net benefit, annualised: EUR 141,168. Planned in the business case: EUR 204,672.
How many of the five targets were met?
Apply
3. Watch the automated steps
What you do here: judge the automated and AI steps by the same standard as everything else.
Quarter after rollout.
| Step | Measure | Result |
|---|---|---|
| Fast-lane check | Small variants cleared without engineer change | 81 % |
| Fast-lane check | Results corrected by an engineer | 7 % |
| AI-assisted order entry | Orders corrected by the clerk, month 1 → month 3 | 12 % → 5 % |
| Milestone messages | Status enquiries per quarter | 412 → 160 |
Engineers correct 7 % of fast-lane results. Good or bad?
Apply
4. Tell signal from noise
What you do here: read a control chart and decide which movements need action.
Every number moves a little from week to week. A control chart draws the normal range from the data itself: the centre line (CL) and upper and lower control limits (UCL, LCL). React to signals, not to every dip. Two common signals: a point outside the limits, or eight points in a row on one side of the centre line.
The limits come from weeks 1 to 8. Now read the weeks after them.
On-time delivery per week, in percent.
| Weeks | Values |
|---|---|
| 1 to 8 | 88, 90, 81, 89, 91, 88, 90, 87 |
| 9 to 16 | 86, 85, 87, 84, 86, 85, 83, 84 |
Week 3 dropped to 81 %. The head of Sales wants an investigation. Do you agree?
Apply
5. Decide what happens next
What you do here: decide whether to sustain, adjust or roll back.
Three of five targets met, 69 % of the benefit realised, the automated steps within limits, and a downward drift in on-time delivery since week 9.
What do you recommend to the process owner?
The two open actions are where the next module starts.
Reflect
Check your understanding
Three questions on the module.
What makes a KPI definition complete?
Take away
What to keep
- A KPI is measure, aggregation and target. Pair speed with quality.
- Compare baseline, target and result, and realised against planned benefit.
- Judge automated steps by end-to-end results, not by usage.
- React to signals, not to noise: limits and runs on a control chart.
- Sustain what works, and turn misses into the next cycle.
After this module you can
- Write complete KPI definitions and pair speed with quality
- Compare results with targets and benefits with the business case
- Judge automated and AI steps by their end-to-end effect
- Read a control chart and react only to signals
- Decide whether to sustain, adjust or roll back