🎯 Quick Answer
What: A reaction plan says what to do when a control chart shows a signal. It is part of the control plan.
The four parts: Set aside the product made since the last good point. Check the likely causes, in order. If the checks find nothing, call the person named for that shift. Write what you found next to the point on the chart.
The test: Someone on the night shift, who was not on the project, can start it alone.
The payoff: The cause is found in the first hour, not the next morning.
Two weeks ago, the process owner put the metric on a one-page control plan. Last week's article showed how to read that metric on a control chart, and the three signals that a process has changed.
This article is about what happens next. A point has just gone outside the control limits. Someone on shift is looking at it. What do they do?
The answer is the reaction plan, the fourth line of the control plan. In DMAIC, the step-by-step method used in Lean Six Sigma, it belongs to Control, the last phase. Here is what a reaction plan is, its four parts, two examples, a template to copy and three common problems.
Why "investigate" is not a reaction plan
Many reaction plans say one thing: investigate. That word asks the person on shift to work out the cause alone.
Here is an example. A coffee plant weighed a sample of bags from its filling line every hour. It plotted the average weight on a control chart.
One night, the point went below the lower control limit. The bags were coming out light. The reaction plan said: "Investigate and correct."
The operator on the night shift was not sure where to start. So the operator turned up the filler setting. The next point went above the upper control limit.
In the morning, the process owner found the cause in a few minutes. The hopper above the filler had been running low, and a low hopper makes light bags.
Only the process owner knew that. The night shift did not. Turning up the filler did not fix the cause. It moved the weight the other way, and the hopper was still low.
A reaction plan puts what the process owner knows on the page, for whoever is on shift.
What a reaction plan is
A reaction plan is a short set of steps for the person who sees a signal on the control chart. It starts when the chart shows any of the three signals: one point outside the control limits, eight points in a row on one side of the center line, or six points in a row going up or down.
It is written for the person on shift, not for the project team. That person may work nights. They may have joined after the project closed. They should be able to start the reaction plan alone, with the tools at the line.
The reaction plan is the fourth line of the control plan. Many teams write a short version on the control plan's page and post the full steps next to the chart.
The four parts of a reaction plan
1. The product
Set aside what was made since the last good point, the last point that was inside the control limits. Keep it apart until someone checks it. This step is called containment. It keeps product that may be wrong away from the customer while the cause is found.
Say exactly what to do: where to put the product, how to mark it, and who decides whether it can ship.
2. The first checks, in order
List two or three likely causes, in the order to check them. Write each one as a plain action the person on shift can do: "check the hopper level", not "investigate the filler".
Put first the cause that comes up most often, or the one that is quickest to check. The project's cause-and-effect analysis is a good place to find the likely causes. Once the plan is in use, the notes on the chart show which causes come up most.
3. Who to call, and when
Name one person to call on each shift, with a phone number. Not a team, and not a person who only works days.
Set a time limit. For example: if the checks do not find the cause within 30 minutes, call. If the product is outside the specification, also say when to stop the line.
4. The record
Write what you found and what you did next to the point on the chart. A few words are enough: "hopper low, refilled".
The notes help in two ways. The next person who sees a signal can read them. And over a few months, they show which causes come up most, so the checks can be put in the right order.
If the checks find nothing
Sometimes the first checks find nothing. Then:
Keep the product set aside.
Call the person named for that shift.
Do not adjust the process to make the next point look better. An adjustment without a known cause hides the signal and adds variation.
Keep plotting, and write "cause not found" next to the point.
If the cause is still not found, or the same signal keeps coming back, the process owner starts a root cause analysis.
An example: the coffee plant's reaction plan
After that night, the coffee plant's process owner rewrote the reaction plan as plain steps. Here it is.
Metric: the average weight of a sample of five bags, weighed every hour.
Starts when: the control chart shows a signal.
The product: set aside the bags filled since the last good sample. Tag them and move them to the hold area. The shift lead decides whether they can ship.
First checks, in order: 1. Check the hopper level. If it is low, refill it and weigh five bags. 2. Check the scale with its test weight. 3. Check the filler setting against the setting sheet.
Who to call: if none of these is the cause within 30 minutes, call the shift lead for that shift. The names and phone numbers are on the sheet.
The record: write what you found and what you did next to the point on the chart.
The plan fits on half a page. The operator on the night shift can start it without calling anyone first.
A second example: the note on the chart
A sawmill measured the thickness of its boards every hour and plotted it on a control chart. One afternoon, a point went above the upper control limit.
The operator followed the reaction plan. The first check, the saw blade, was fine. The second check found the cause: a saw guide was loose. The operator tightened it and wrote "Saw guide loose" next to the point on the chart.
Three months later, the same signal showed up on a different shift. That operator had not been there the first time. But the operator read the note, checked the saw guide first, and found it loose again.
The process owner, the mill supervisor, moved the saw guide to the first check. And because the same cause had come up twice, the maintenance team looked at why the guide kept coming loose.
A reaction plan template
Copy this onto the control plan, or onto a sheet next to the chart.
Metric and chart: which metric, how often it is measured, and where the chart is posted.
Starts when: the control chart shows a signal.
The product: what to set aside, where to put it, how to mark it, and who decides whether it can ship.
First checks, in order: 1. The most likely cause, as one plain action. 2. The next most likely. 3. The third.
Who to call: one name and phone number for each shift, and the time limit.
If the checks find nothing: keep the product set aside, call, do not adjust, keep plotting.
The record: write what you found and what you did next to the point on the chart.
Process owner: the person who keeps this reaction plan up to date.
Three common problems
"Investigate" is the only step. It asks the person on shift to work out the cause alone, often at night. Replace it with two or three plain checks, in order.
The person to call only works days. A plant that runs three shifts needs a name for each shift. If nobody is named, the night shift writes a note and waits for the morning.
The reaction plan never changes. The notes on the chart show new causes, and causes that come up again and again. When a cause comes up twice, move it up the list, and ask why it keeps happening.
What leaders can do
Ask an operator what they would do first if the next point went outside the control limits. Ask on a night shift too, not only on a day shift.
Read the notes on the chart in the monthly review, not only the points.
When a signal shows, ask what the checks found, and whether the reaction plan needs to change.
Make sure the person named on each shift can act without waiting for approval.
I write one of these every week, based on what actually happens on projects rather than what the textbook says should happen. You can join the newsletter here.
FAQ
What is a reaction plan?
A reaction plan is a short set of steps that says what to do when a control chart shows a signal. It says what to do with the product made since the last good point, which causes to check and in what order, who to call, and what to write on the chart. It is part of the control plan.
What should a reaction plan include?
Four parts. The product: what to set aside and who decides whether it can ship. The first checks: two or three likely causes, in order, each a plain action. Who to call: one name for each shift, and a time limit. The record: a note next to the point on the chart.
Is a reaction plan part of the control plan?
Yes. The reaction plan is one line of the control plan. Many teams write a short version on the control plan's page and post the full steps next to the control chart, where the person on shift will see them.
Who writes the reaction plan?
The project team writes the first version with the process owner, before the project closes. The process owner keeps it up to date. Test it with the people who will use it: ask an operator on each shift what they would do first.
What if the checks do not find the cause?
Keep the product set aside, call the person named for that shift, and keep plotting. Do not adjust the process to make the next point look better. If the cause is still not found, or the signal keeps coming back, the process owner starts a root cause analysis.
Write the first checks, in order, and who to call.
