Why we don't compute an RPN
Classical FMEA multiplies three ordinal ratings into one number. We report the count instead.
Ask any reliability engineer about FMEA and within a minute they will ask where the Risk Priority Number is. It is the classical output: severity times occurrence times detectability, each rated one to ten, multiplied into a single score used to rank what to fix first.
We do not compute one, and the reason is arithmetic rather than ideology.
Multiplying ordinal ratings pretends they are interval quantities. A severity of 8 is not twice a severity of 4; the numbers are labels in a rank order, and multiplying labels produces a number with no units and no meaning.
Detectability belongs to your system, not ours
The third term is worse. Detectability asks how likely you are to catch the failure before it reaches a user — which depends entirely on your logging, your evals, your review process and your product surface. It is a property of your deployment, not of the model. Estimating it on your behalf would be inventing a number about a system we have never seen.
So severity axes are published per mode, occurrence is measured with a confidence interval, and detectability is declared out of scope. Where one figure is genuinely wanted, the census reports a count: how many modes an intervention actually reached.
Every figure here describes something measured and committed. See the measurements · read the method