Learning objective
By the end of this lesson, you can distinguish a symptom, causal factor and corrective action, then test whether the action is strong enough.
Why it matters
An investigation can identify a plausible cause and still fail if the corrective action does not change the condition that produced the event.
Core concept: causal link and action link
RCA needs two defensible links:
- Causal link: evidence shows how a condition contributed to the event.
- Action link: the intervention changes that condition and has a measure.
Visual explanation
Worked example
Evidence shows that rescheduling updates one table while reminders read another.
Possible actions:
- email the engineering team to be careful;
- retrain support agents;
- repair and backfill the projection;
- add an automated consistency check before send;
- test all downstream consumers when scheduling events change.
The first two actions do not address the verified condition. The last three change the system at different layers.
Measures:
- implementation: all affected records backfilled; new integration test active;
- outcome: reminder-time mismatch rate below 0.1% for four weeks.
Common mistake
Mistake: using “human error” as the end of analysis.
People act within information, tools, incentives and constraints. Ask what made the action possible and why controls did not prevent or detect it.
Quick check
Which is an outcome measure?
A. Training delivered to 20 staff.
B. New checklist published.
C. Mismatch complaints remain below 0.1% for four weeks.
D. Action owner assigned.
Answer: C. It measures the result, not implementation activity.
Practical prompt
Choose one recurring problem. Write one symptom, one candidate causal factor, evidence that would verify it, one system-level action and one outcome measure.
Summary
- Symptoms are not causes.
- Causal claims need evidence and a mechanism.
- Corrective actions need implementation and outcome measures.
Next lesson
Continue to 5 Whys to deepen one focused causal chain.