Each FMEA row gets three scores from 1 to 10. Severity (S) rates how bad the effect is if it reaches the user. A 1 means the effect is barely noticeable; a 10 means it could harm the patient or user with no warning. Severity is about consequence only. It does not care how often the failure happens or whether you can catch it.
Occurrence (O) rates how often the cause is expected to produce the failure mode with the current process. A 1 means it practically never happens; a 10 means it happens very frequently or almost every time. Base this on real data where you have it: scrap records, complaint logs, test results. Where you have no data, score conservatively and note the assumption.
Detection (D) rates how likely your current controls are to catch the failure mode or its cause before the product leaves the step or the plant. Here the scale runs the opposite way to what people expect: a 1 means your controls almost certainly catch it; a 10 means there is no control or it cannot be detected. A high detection score is bad news.
Standard Procedure
- 1Agree on written one-line descriptions for each score level in your own words before scoring anything.
- 2Score severity first, based on the worst credible effect on the end user.
- 3Score occurrence from actual failure data for the cause; if none exists, estimate and flag it.
- 4Score detection by asking how reliably the current control catches the problem, remembering that 10 means it will not.
Lesson Summary
- ✓Severity describes consequence, occurrence describes frequency, detection describes catch probability.
- ✓Detection is scored so that a strong control gets a low number.