Replacing RPN with AP: Why Prioritizing Actions Became More Confusing? — Five Steps to Implement FMEA Action Priority

After a car parts factory updated its FMEA, it changed the "RPN" column to the "AP" column. The first cross-functional review meeting got stuck: out of 260 failure causes in the entire table, 47 were rated as H, and over 90 were rated as L. The process supervisor asked a question that no one could answer: "Previously, if RPN was greater than 100, we would take action. Now, with these 47 H-rated items, in what order should we address them?" Even more awkward were the 90+ L-rated items: since they have a low priority, can we just ignore them?

The problem is not that the AP table was calculated incorrectly, but that the team treated it as a "new RPN," assuming it was a more advanced risk score. However, AP is never a risk score.

1. Two Fatal Flaws of RPN

RPN is the product of three ratings, which brings two issues.

The first issue is equal weighting. An item with a severity of 9, occurrence of 6, and detection of 1 has an RPN of only 54; whereas a surface scratch with a severity of 3, occurrence of 10, and detection of 7 has an RPN of 210. According to the most popular company rule, "if RPN is greater than 100, take action," resources would be allocated to the surface scratch first, while the safety-related item, with only 54 points, would be left untouched. Multiplication dilutes the severity.

The second issue is the threshold. Each company sets its own thresholds, such as 100, 125, and 80, with no standard basis. Moreover, the same numerical value can come from entirely different combinations: a severity of 10 × occurrence of 2 × detection of 5 and a severity of 4 × occurrence of 5 × detection of 5 both result in 100 points, but the risk implications are completely different.

The 2019 edition of the AIAG-VDA FMEA manual therefore replaced RPN with Action Priority (AP): instead of multiplication, it uses a table to determine H (high), M (medium), and L (low) based on the combination of severity, occurrence, and detection. The weighting order is severity first, occurrence second, and detection last.

2. What the AP Table Says: Three Hidden Rules

First, remember the most easily overlooked statement in the manual: AP is not a classification of high, medium, or low risk, but a priority for "whether action needs to be taken to reduce risk." It answers "which one to do first and whether to do it," not "how big the risk is."

These three rules are most useful when reading the table.

Rule One: Severity Takes Priority. Items with a severity of 9-10 and an occurrence of 6-10 are always rated as H, regardless of the detection score. Items with a severity of 9-10, an occurrence of 4-5, and a detection of 7-10 are also rated as H. Conversely, items with a severity of 9-10 but an occurrence of only 2-3 and strong detection capability (detection of 2-4) are rated as L. This might seem counterintuitive, but the logic is solid: if the probability of occurrence is extremely low and can be intercepted, there is no need to allocate resources at the moment.

Rule Two: Occurrence is the Second Priority. Items with a moderate severity (4-6), high occurrence (8-10), and poor detection capability (detection of 5 or more) are rated as H—frequent defects will eventually slip through.

Rule Three: Detection Has the Lowest Weight. For example, an item with a severity of 4-6 and an occurrence of 8-10, if the detection score is improved from 5 to 3, the AP only drops from H to M; but if the occurrence is reduced from 8 to 3, it often falls directly to L. The true way to bring down an H rating is through preventive measures that reduce occurrence, not by adding inspections. This is the opposite of the old RPN approach, where "adding inspections to lower the score" was a common practice.

Additionally, the manual explicitly recommends: items with a severity of 9-10 and an AP of H or M should at least be submitted for management review. This rule should be written into the company's procedures; otherwise, these items will never appear in the review meetings.

3. Five Steps to Implementation

Step One: Divide into Three Piles and Fix the Handling Methods for H and L. Recalculate the AP for the entire table and handle items according to the conclusions: H items must have specific actions or a written justification for why the current controls are sufficient; M items are included in the improvement plan; L items are not mandatory for action, but the reasons for not taking action must be documented. For example, the 90+ L items should have evidence-based reasons, such as "appearance-related defects, not listed as special characteristics by the customer, existing visual inspection capability verified by MSA, detection rated as 5." If no reason can be provided for an L item, it indicates a problem with the rating or control description, and it should be sent back for re-evaluation.

Step Two: Reorder Within the H Items. With only three levels of AP, the 47 H items cannot be prioritized using AP alone. Three supplementary rules are needed: prioritize items with higher severity (9-10 over 7-8); among items with the same severity, prioritize those with higher occurrence; within the same type of action, prioritize preventive measures that reduce occurrence over measures that improve detection. After reordering, assign a number to each item, which will determine the order of resource allocation and the agenda for the review meeting.

Step Three: Assign a Responsible Person, Completion Time, and Verification Method for Each H Item. An FMEA with only a description of the action is equivalent to having no action. It is recommended to tie the closure of H items to the APQP stage gates or before PPAP submission, and M items to the next FMEA review. When a customer audit asks, "When will this item be closed?" being able to provide a date and responsible person is essential for a closed loop.

Step Four: Reassess Occurrence and Detection After Actions are Taken. This is the most frequently overlooked step. AP only represents task allocation, not the elimination of risk. After implementing the actions, reassess the occurrence (whether prevention is effective) and detection (whether detection has improved), and recalculate the AP. Document the version history in the change record. Simply checking the action column and leaving the ratings unchanged will result in the old risk being seen in the next review.

Step Five: Integrate AP into the Control Plan and Special Characteristics List. The failure causes corresponding to H items must have corresponding preventive and detection controls in the control plan and corresponding poka-yoke or inspection requirements in the work instructions. Causes with a severity of 9-10 should be reflected in the special characteristics list, triggering the corresponding capability and monitoring requirements. Mismatch between the actions listed in the FMEA and the documents used on-site is the most common reason for nonconformities during customer audits.

4. An Example

Two failure causes on the same stamping process.

The first cause is improper forming leading to misalignment of installation holes, with safety implications (severity 9), frequent occurrence (occurrence 6), and 100% dedicated inspection at the end of the line, verified by MSA (detection 1). The RPN is 54 points, which does not even meet the threshold under the old rule. Checking the AP table: severity 9-10, occurrence 6-7, rated as H—action must be taken, or a written explanation of why the current controls are sufficient must be provided.

The second cause is surface scratches, which the customer considers only a perception issue (severity 3), with a high occurrence (occurrence 10) and poor detection (detection 7). The RPN is 210 points, ranking it at the top of the entire table under the old rule. Checking the AP table, it is rated as M.

An item with 54 points is rated as H, while an item with 210 points is rated as M, completely reversing the order. The correct approach is to allocate resources to the first item first: add a positioning structure to the mold to prevent misalignment, reducing the occurrence from 6 to 2, and thus lowering the AP. The second item should be recorded in the M list, with a documented basis for not allocating resources, and left for improvement when the opportunity arises. In the RPN era, this was impossible because it could not compare "safety" and "scratches" on the same scale.

5. Three Common Pitfalls

Pitfall One: Treating AP as a Risk Score for Prioritization. Ranking H items by occurrence in descending order and ignoring severity is equivalent to discarding Rule One.

Pitfall Two: Treating L as "No Need to Look." L items are not mandatory for action, but they require a basis for judgment. L items without a reason will not hold up in the field and will also deprive the team of an opportunity to verify the correctness of the ratings.

Pitfall Three: Only Updating the FMEA Table, Not the Downstream Documents. If the control plan, work instructions, and special characteristics list are not updated accordingly, the risk remains on the shop floor.

6. One Sentence Summary

RPN calculates how big the risk is, while AP determines which action to take first—using AP as a risk score will only make the prioritization more confusing.

AP is not a risk classification but an action priority: H items must be addressed or have a written justification, M items should be addressed, and L items also require a basis for judgment.

Knowledge code: 8.3.1

Version: v20260924

Author: QTank QTank is dedicated to providing systematic professional knowledge, methodologies, and practical tools to quality management practitioners, helping companies continuously improve their quality capabilities.