RPN+ against Action Priority
The AIAG & VDA handbook retired RPN in 2019 for the same reason set out above, and replaced it with Action Priority. AP and RPN+ diagnose the same disease and prescribe differently. Here is the comparison, including where AP wins.
| Action Priority | RPN+ / RE+ | |
|---|---|---|
| What it returns | Three categories: high, medium, low | 1000 distinct values, or 25 without history |
| Ordering inside a priority | None. Forty high items arrive unranked | Total order, no ties anywhere |
| Severity dominance | Yes, built into the table | Yes, built into the elegant mathematics |
| Basis | A published lookup of every rating combination | One line of elegant mathematics, checkable by hand |
| Rating scales | Only valid against the handbook's own scales | Any 1–10 scale, including one you wrote |
| Score traces to its ratings | No. "High" does not say which factor drove it | Yes. 754 is impact 7, incidence 5, detection 4 |
| When occurrence cannot be counted | You can guess, you enter the Speculation Realm, the auditor can sign off without the occurrence rating. | Switches to RE+ and records that it did |
| Recognised by OEMs | Yes — the automotive standard | No, it is elegant mathematics: RPN+ and RE+ that are injective mixed-radix encodings that induce a lexicographic priority order. Definitely, not a standard. |
| Audit position | Expected. Its absence is a finding | Supplementary. Report it alongside |
| Cost to adopt | Licensed handbook, new scales, retraining | None. Same ratings, same tables |
The two cases from the top of this page, and a third
The handbook's own rationale for retiring RPN uses the first pair, so Action Priority separates them by design. The interesting case is the second block, where three modes share a severity and any severity-driven banding must place them together.
| Ratings | RPN | AP | RPN+ | |
|---|---|---|---|---|
| Frequent, trivial | I1 N10 D1 | 10 | separates | 201 |
| Rare, catastrophic | I10 N1 D1 | 10 | separates | 1011 |
| Injury, rare, caught in-house | I9 N2 D2 | 36 | same band | 922 |
| Injury, rare, escapes to the customer | I9 N2 D9 | 162 | same band | 929 |
| Injury, occasional, caught | I9 N6 D2 | 108 | same band | 962 |
Three modes that all injure somebody. AP places them together and stops — it has told you to act, which is its job. Now decide which one to start on Monday.
RPN answers, and answers wrongly. It puts the escaping one first at 162, ahead of the one that happens three times as often at 108 — because a bad detection rating multiplies harder than a real frequency does. RPN+ orders them 962, 929, 922: most frequent first, then the one that escapes, then the one already caught. Incidence outranks detection because it sits in the higher column, which is the order you would have chosen anyway.
Read that table honestly and AP wins the argument that matters most in automotive. If your customer asks for Action Priority, you give them Action Priority — nothing here replaces it, and this tool does not pretend to. RPN+ is not a standard and will not satisfy an auditor who came looking for one.
Where RPN+ earns its place is the row AP leaves blank: ordering inside a priority. AP tells you that forty modes need action. It cannot tell you which to start on Monday, and in practice teams fall back to sorting by RPN — which reintroduces every flaw AP was created to remove. RPN+ is what that fallback should be. Use AP to decide whether to act; use RPN+ to decide the order. They answer different questions and neither answers both.
The second row worth noticing is the last but one. AP assumes an occurrence rating exists. On a new design it does not, and the AP handbook has no mechanism to record that the number was estimated rather than counted. That gap is the one this method was built around, and it is why the fork and RPN+/RE+ belong together rather than as two separate ideas.