Prevention vs Detection Controls in FMEA: Which One Reduces Risk Faster

The two columns for current controls in a PFMEA column set do different things, reduce different ratings, and cost different amounts to implement. Most FMEA training covers the definitions; fewer discussions address the strategic question: when a team has budget for one improvement, which type of control reduces risk more effectively? This post covers the mechanics, the rating impact, and the practical investment logic for prevention versus detection in manufacturing FMEA.

What Each Control Type Does

A prevention control reduces the likelihood that a failure cause occurs in the first place. It targets the Occurrence rating. Prevention controls act upstream of the failure mode—they reduce the frequency at which the cause can trigger the failure at all. Examples: process capability improvement (reducing tool wear variation through scheduled replacement intervals), error-proofing fixtures that physically prevent part misloading, validated torque control that eliminates operator-to-operator variation in fastener tightening, and SPC with control limits that trigger intervention before parts go out of spec.

A detection control identifies a failure mode or its cause after the failure has occurred but before the part or assembly reaches the customer. It targets the Detection rating. Detection controls act downstream of the failure mode—they catch what has already gone wrong. Examples: dimensional inspection at end-of-line, in-process CMM measurement, functional test at assembly, vision systems checking for missing features, and automated torque verification after joining.

The timing distinction is the critical one: prevention stops the failure; detection finds it after it happens. A detection control with a rating of 1 (nearly certain to detect) means you are almost certainly catching every defect—but those defects still occurred. The occurrence rating, and the cost of producing nonconforming parts, doesn’t change.

How Each Type Moves the Risk Numbers

Prevention and detection affect different axes of the risk assessment, which matters whether your organization uses RPN or AIAG-VDA Action Priority.

Adding a prevention control lowers the Occurrence rating. Since RPN = S × O × D, a lower O directly reduces RPN. Under Action Priority, lower Occurrence can move an item from High to Medium priority when Severity is below 9—though for severity 9–10, AP remains High regardless of Occurrence.

Adding a detection control lowers the Detection rating. This also reduces RPN directly. However, under AIAG-VDA Action Priority, the AP table weights Detection less heavily than Occurrence for high-severity items. A failure chain with severity 8 and occurrence 4 carries different AP implications depending on detection, but two failure chains—one with O=6 and one with O=2—are not equivalent risks even if their RPNs are similar. The AIAG-VDA methodology was designed specifically to correct for this RPN artifact.

AIAG-VDA’s Explicit Preference for Prevention

The AIAG-VDA FMEA Handbook (1st Edition 2019) is direct on this point: when recommending actions, teams should prioritize error-proofing (prevention) over detection improvement. The recommended hierarchy for Optimization (Step 6) is:

1. Error-proofing (poka-yoke) — makes the failure cause physically impossible or immediately self-evident
2. Process capability improvement — reduces the frequency of the cause through statistical control
3. Enhanced detection — improves the ability to catch failures that do occur

This ordering reflects a real quality cost difference. A detection control that catches 100% of defects still means the defects were produced—you’re spending resources on inspection, rework, and scrap. A prevention control that eliminates the defect eliminates all downstream costs simultaneously.

The hierarchy also reflects failure risk. Detection controls can fail: a calibration drift, a false accept decision, a process deviation that produces defects the detection method misses. Prevention controls that physically prevent a failure cause from occurring don’t have a “detection failure rate” to manage.

Practical Examples of Each Type

The distinction becomes clearer with side-by-side examples from the same failure chain.

Failure mode: Fastener installed with insufficient torque, causing joint separation in service.

The prevention options eliminate the problem at the source. The detection options find it later, after the defect has already been produced. Both serve a role—but a team that relies solely on end-of-line inspection has documented that defects occur at some rate and that detection is the only safeguard.

When Detection Is the Right Investment

There are cases where investing in detection rather than prevention is the correct decision:

Severity 9–10 with occurrence already reduced by prevention. Once occurrence is as low as practical, adding a detection layer provides defense-in-depth for the highest-severity failure modes. For safety-critical characteristics, redundant controls—both prevention and detection—are common practice.

Failure modes where prevention is not technically feasible. Some causes cannot be eliminated through process control. Raw material variation from supplier processes, tool wear curves that aren’t predictable enough for scheduled replacement, and environmental factors outside the process window all fall here. Detection becomes the primary safeguard when prevention has reached its practical limit.

New or unstable processes where occurrence data doesn’t yet exist. Early production runs on new tooling often benefit from robust detection first—to understand actual defect patterns—before permanent prevention controls are designed. The detection data informs what the prevention investment should target. This is a temporary strategy, not a permanent risk management approach.

How to Document the Control Type in the FMEA

AIAG-VDA format uses separate columns for prevention and detection controls. A common documentation error is listing detection controls in the prevention column (or vice versa) because the distinction wasn’t considered at write time. This matters during audit: an auditor who finds “100% inspection” in the prevention column will flag it as a methodology error.

Each control should be assigned to the column that matches what it actually does:

• If the control makes the failure cause less likely to occur → Prevention Controls column
• If the control identifies the failure or cause after it has occurred → Detection Controls column

Controls that do both (a smart tool that both enforces correct torque AND flags a failed joint) should be split: list the torque enforcement in Prevention, list the failed-joint flag in Detection. Don’t bundle them in one entry—they affect different ratings.

For more on how to evaluate your existing detection controls and assign detection ratings consistently, the post on FMEA detection ratings and evaluating current controls walks through the 1–10 rating criteria with manufacturing examples. When prioritizing which control improvements to invest in given a fixed budget, the action prioritization guide covers the AP matrix approach for sequencing work across multiple open items.

The RPN and action priority calculator lets you model the impact of a specific control improvement before committing to it: lower the Occurrence rating by 2 points to reflect a prevention improvement, or lower Detection by 2 points to reflect a detection upgrade, and see how the AP or RPN changes. Comparing both scenarios side by side shows whether your planned investment reduces risk more through prevention or detection—and which path gets you off the High priority tier faster.