Stage 4 · Sustain · 09 · 20 min
Find the next improvement and repeat
A process that is not managed slides back. This module makes sure Northstar’s does not, and finds what to improve next.
The project is over; the process goes on. You set up the control plan, the learning loop and the owner’s review, then follow the bottleneck to where it moved and pick the next improvement.
Apply
1. Write the control plan
What you do here: decide what happens when each monitored number drifts.
A control plan names what is measured, how often, by whom, the limit, and the reaction plan. Without a reaction plan, monitoring is just watching.
Which reaction belongs to each measure?
On-time delivery, weekly control chart: a run of eight below the centre line
Orders waiting for engineering review above 6
Engineers correct more than 10 % of fast-lane results
First-pass yield below 90 % in a month
Apply
2. Close the learning loop
What you do here: put the steps in order that turn each correction into a better process.
Every time a person corrects the process, whether a clerk fixing an order or an engineer overriding the fast lane, the process can learn something. It only does if the correction travels somewhere.
Click the steps in order.
What comes first?
The same loop updates the work instructions people use. Test cases, rules and instructions improve together.
Apply
3. Run the owner’s review
What you do here: set up the monthly process review so it produces decisions.
The first monthly process review: department heads present twelve slides each about their part of the process. After 90 minutes, everyone agrees things are going well.
What do you change for next month?
Apply
4. Follow the bottleneck
What you do here: find where the bottleneck moved and log it as the next issue.
Typical custom order, days waiting before each step, before and after the redesign.
| Before step | Before | After |
|---|---|---|
| Engineering review | 4 | 0.7 |
| Schedule production | 2 | 2 |
| Manufacture | 3.5 | 3.5 |
| Final inspection | 2.5 | 3 |
| Pack and ship | 1 | 1 |
Which wait got longer after the redesign?
Log it: “Finished parts wait for the weekly inspection batch; cause: batch inspection, made worse by faster upstream flow.” It goes into the register next to the remaining rework, and both are priced with the method from module 5.
Apply
5. Pick the next improvement
What you do here: re-score the candidates from module 1 and choose where the next cycle starts.
The leadership team re-scored the candidates after the project. Dysfunction 5 means badly broken.
| Process | Importance | Dysfunction | What changed |
|---|---|---|---|
| Engineer and fulfil custom order | 5 | 3 | On time 86 %, but rework and the inspection batch remain |
| Fulfil catalogue and repeat order | 4 | 2 | Unchanged, works well |
| Issue to Resolution | 3 | 2 | Status enquiries down from 412 to 160 |
| Procure to Pay | 2 | 3 | Unchanged |
What does the custom order process score now (importance × dysfunction)?
The cycle starts again at module 3 for the same process, with the register, baseline and owners already in place. That is why the second cycle is faster than the first.
Reflect
Check your understanding
Three questions on the module.
What turns monitoring into control?
Take away
What to keep
- A control plan needs limits, reactions and owners.
- Turn every correction into a test case and a better rule.
- The owner’s review decides; it does not just report.
- After every improvement, look for where the bottleneck moved.
- Re-score the landscape and start the next cycle.
After this module you can
- Write a control plan with limits, reactions and owners
- Turn corrections into test cases and better rules
- Run the process owner's review as a decision forum
- Find where the bottleneck moved after an improvement
- Re-score the landscape and choose the next cycle