Change Point Analysis is a problem-solving method that looks for differences between the period before a problem appeared and the period after it began.
The basic question is:
What changed when the problem changed?
The method is useful for problems that started suddenly or became significantly worse at a known point in time.
Define when the problem changed
Start with the timeline.
Ask:
- When was the process last known to be normal?
- When was the first known abnormal result?
- Did the problem begin suddenly or gradually?
- Was the change tied to a shift, lot, product, machine, or event?
5W1H Problem Definition helps establish the timing, location, scope, and observed condition.
The narrower the change window, the easier it is to investigate.
Compare before and after conditions
Look for changes in categories such as:
- material;
- machine;
- method;
- measurement;
- people;
- environment;
- software;
- supplier;
- schedule.
Examples include:
- new material lot;
- maintenance intervention;
- parameter change;
- new operator;
- revised work instruction;
- tooling replacement;
- software update;
- environmental shift.
4M Analysis can provide a useful structure for organizing possible change points.
Use evidence, not memory alone
Useful sources include:
- change records;
- maintenance work orders;
- production history;
- quality data;
- supplier records;
- parameter logs;
- training records;
- shift handovers;
- software release history.
People may remember significant events, but records help confirm timing.
Distinguish change from cause
A change that occurred near the problem is a candidate cause, not proof.
For example:
Tooling was changed one hour before defects appeared.
The team should then ask:
- Could this change physically create the defect?
- Is the timing consistent?
- Does the problem disappear if the condition is restored?
- Does the same change reproduce the problem?
Root Cause Analysis should continue until the causal mechanism is verified.
Compare good and bad conditions
If some units are good and others bad, compare them.
Look for differences in:
- lot;
- machine;
- cavity;
- operator;
- shift;
- setting;
- supplier;
- time.
Is/Is Not Analysis can help structure these contrasts and narrow the investigation.
Use the method for intermittent problems
Change Point Analysis can also help when a problem comes and goes.
The team can compare periods when the condition is present with periods when it is absent.
This may reveal recurring triggers such as:
- temperature;
- product mix;
- staffing;
- maintenance activity;
- material source.
Connect change control to prevention
When the cause is linked to an uncontrolled process change, the corrective action may include stronger change management.
Management of Change can help control technical or operational changes before they are released into normal operation.
Common mistakes
Searching every historical change without narrowing the problem window, treating coincidence as proof, relying only on memory, ignoring software or supplier changes, failing to compare good and bad conditions, and stopping at “something changed” without verifying the causal mechanism are common mistakes.
Practical sequence
- define the abnormal condition.
- identify when the problem began or worsened.
- establish the last known good condition.
- list changes inside the relevant window.
- compare good and bad conditions.
- prioritize changes that could physically explain the effect.
- verify timing and evidence.
- test the suspected mechanism.
- implement corrective action.
- improve change controls where appropriate.
The practical lesson
Change Point Analysis turns the question “Why did this suddenly happen?” into a disciplined comparison of conditions.
Finding what changed can dramatically narrow a root cause investigation.