Most manufacturing FMEA registers end up as a spreadsheet built once for an audit and never opened again — hundreds of failure modes scored, ranked, and then quietly ignored the moment the compliance deadline passes. That's not a documentation problem, it's a workflow problem: the scores were never connected to anything a maintenance team actually does every day. iFactory's FMEA-to-maintenance engine links every high-risk failure mode directly to a PM task, inspection route, or condition monitoring trigger — so the analysis drives real work instead of sitting in a folder.
Three Numbers Decide Which Failure Gets Fixed First. Most Plants Score Them Wrong.
Severity, Occurrence, and Detection — each rated 1 to 10 — multiply into the Risk Priority Number that should drive your entire preventive maintenance strategy. Get the scoring discipline wrong and the RPN ranks cosmetic issues above safety-critical ones.
What FMEA Actually Asks, In Plain Terms
Failure Mode and Effects Analysis is a structured way of asking three questions about every component in a system, one at a time, before anything actually breaks: What can fail? What happens when it does? How would we know before or when it does? The discipline is in doing this systematically across an entire asset or process, rather than only thinking about failure modes after something has already gone wrong on the floor.
Failure Mode
The specific way a component can fail — bearing seizure, seal leak, voltage spike, misalignment, contamination ingress. Not a vague category like "mechanical problem," but the precise mechanism.
Effect
What happens to the system, the product, or the operator when that specific failure mode occurs — line stoppage, scrap, safety exposure, or a downstream failure it triggers in another component.
Cause
The underlying mechanism that produces the failure mode — fatigue, corrosion, contamination, overload, or a design or installation flaw that makes the failure more likely to occur.
Current Controls
Whatever is already in place to prevent or detect the failure — a PM task, an inspection interval, an alarm, or a redundant system. Many teams discover during FMEA that assumed controls don't actually exist.
Scoring Severity, Occurrence, and Detection Without Guessing
The single biggest source of a broken FMEA register is inconsistent scoring — one engineer's "7" is another engineer's "4," and the resulting RPN ranking becomes meaningless the moment different people score different assets without a shared reference. A usable scoring guide anchors every number to a concrete, observable definition.
| Score | Severity (S) | Occurrence (O) | Detection (D) |
|---|---|---|---|
| 9-10 | Safety hazard or complete line shutdown with no workaround | Failure has occurred multiple times per month historically | No current control detects the failure before it causes the effect |
| 7-8 | Major production loss, significant scrap, or regulatory exposure | Failure occurs roughly monthly based on work order history | Control exists but has a low historical catch rate |
| 4-6 | Moderate disruption, contained to one line or process step | Failure occurs a few times per year | Control catches the failure most of the time, with occasional misses |
| 1-3 | Minor or cosmetic effect with negligible operational impact | Failure has occurred rarely or never in available history | Control reliably detects the failure before it causes the effect |
Anchor every score to actual work order history and documented incident data wherever it exists, rather than relying purely on an engineer's memory or intuition. A failure mode scored as Occurrence 3 because "it doesn't happen that often" should be checked against the CMMS — if the work order history shows six occurrences in the past year, the score is wrong regardless of how infrequent it feels to the team scoring it.
An FMEA Register Is Only as Good as the Data Behind Each Score
iFactory pulls actual failure frequency and downtime history directly into the scoring process, so Occurrence and Severity ratings are grounded in your CMMS data instead of guesswork.
RPN vs. Action Priority: Why the Classic Formula Has a Blind Spot
The traditional RPN calculation treats all three factors as equally weighted, multiplying them together into a single number from 1 to 1,000. This produces a well-known distortion: a failure mode scored Severity 10, Occurrence 1, Detection 1 produces an RPN of 10 — identical to a failure mode scored Severity 1, Occurrence 10, Detection 1. One of those is a rare but catastrophic safety risk. The other is a frequent but cosmetic annoyance. A flat RPN threshold cannot tell them apart, which is exactly the failure mode the newer AIAG-VDA Action Priority method was built to correct.
The FMEA Process, Step by Step
Define the scope
Select the asset, system, or process step being analyzed. A full-plant FMEA attempted in one pass overwhelms most teams — scope it to a single critical line or asset class for the first cycle and expand from there.
List every credible failure mode
Break the asset into components and systematically list how each one can fail — not just the failure modes the team already knows about, but ones surfaced by reviewing work order history, OEM documentation, and similar-asset failure patterns.
Document effects, causes, and current controls
For each failure mode, record what happens when it occurs, what mechanism drives it, and what's already in place to catch or prevent it. This is the stage where assumed controls are frequently discovered not to exist in practice.
Score S, O, and D against the anchored scale
Rate each dimension 1-10 using a shared, documented scoring guide, cross-checked against actual failure history wherever it's available rather than relying purely on memory or intuition.
Calculate RPN and apply Action Priority
Multiply S × O × D for the RPN, but cross-reference against an Action Priority lookup so high-severity, low-frequency failures aren't buried beneath frequent, low-consequence ones on the final ranked list.
Assign corrective actions with owners and dates
For every failure mode above the risk threshold, and every high-severity item regardless of RPN, define a specific corrective action, a named owner, and a target completion date — then re-score after implementation to verify the risk actually dropped.
A Worked Example: Scoring Three Failure Modes on One Conveyor
Consider a single transfer conveyor feeding a packaging line, analyzed for three of its most credible failure modes. The scoring illustrates exactly why a flat RPN cutoff can mislead a team that doesn't also check Action Priority.
| Failure Mode | S | O | D | RPN | Action Priority |
|---|---|---|---|---|---|
| Drive belt tracking failure — minor product scuffing | 2 | 7 | 3 | 42 | Medium |
| Guard interlock bypass — exposed pinch point | 10 | 2 | 4 | 80 | High |
| Motor bearing wear — gradual, well-monitored | 5 | 6 | 2 | 60 | Low |
Ranked by RPN alone, the guard interlock failure at 80 would appear as the second priority behind the bearing wear item, and only narrowly ahead of the belt tracking issue. But the interlock bypass carries a Severity of 10 — a genuine safety hazard — and under Action Priority logic it's flagged High regardless of its comparatively low RPN, correctly outranking both other items. This is the exact distortion the classic formula introduces, and exactly why a mature FMEA program cross-checks Action Priority rather than sorting a spreadsheet by RPN and stopping there.
A Composite Scenario: What a Rushed FMEA Missed
A mid-size food processing plant ran its first FMEA on a critical filling line ahead of an upcoming customer audit, under a tight two-week deadline. The team — a single quality engineer working largely alone, without input from the maintenance technicians who actually serviced the equipment — scored forty-two failure modes across the line's major components in a series of desk-based sessions, relying mostly on memory and the equipment manuals rather than work order history.
The register looked thorough on paper and satisfied the audit. But six months later, a filler nozzle seal failure caused a two-hour unplanned stoppage and a batch of product that had to be scrapped for contamination risk. When the reliability team went back to the FMEA register afterward, they found the seal failure mode had been scored Occurrence 2 — "rarely occurs" — based on the quality engineer's general impression of the equipment. The actual work order history, which nobody had cross-checked during the original scoring session, showed the same seal had already failed twice in the eighteen months prior. Properly scored against that history, Occurrence should have been closer to 6, which would have pushed the failure mode's RPN well above the threshold that triggers a corrective action — likely a shortened seal replacement interval that would have caught the failure before it caused the stoppage.
The plant's response was not to abandon FMEA, but to change how it was conducted going forward. Subsequent FMEA cycles pulled CMMS work order history for every component before scoring began, and included a maintenance technician and a line operator in every scoring session alongside the quality engineer. The corrected process did not just fix the seal failure mode — it surfaced several other Occurrence scores across the register that had been quietly underestimated the same way, each one now grounded in actual failure data instead of general impression.
Connecting FMEA Output to Actual Maintenance Work
An FMEA register that ranks failure modes but never changes what the maintenance team actually does is analysis for its own sake. The value of the exercise is entirely in the corrective actions it generates and whether those actions are tracked through to completion and verification.
High RPN, Detectable Precursor
When a high-scoring failure mode has a detectable warning sign with useful lead time — vibration change, temperature drift, pressure deviation — the corrective action is a condition monitoring trigger tied to that specific signal, not a generic schedule change.
High RPN, No Detectable Precursor
When no useful early warning signal exists, the corrective action is typically a revised time-based PM interval, a design change to reduce the failure's likelihood, or pre-positioning spare parts to shorten repair time when the failure does occur.
High Severity Regardless of RPN
Any failure mode scored Severity 9 or 10 needs a corrective action regardless of how the RPN or Action Priority calculation comes out — the consequence is too severe to leave unaddressed even if the occurrence is genuinely rare.
Low Priority, Monitor Only
Failure modes that score low across all three dimensions don't need a new corrective action, but they should stay in the register and be re-checked at the next review cycle in case operating conditions or failure history changes.
Each corrective action needs a named owner and a target completion date recorded directly in the register, not left as an informal understanding. Once implemented, the affected failure mode should be re-scored — typically Detection improves if a monitoring trigger was added, or Occurrence improves if a design or procedural change addressed the root cause — to confirm the corrective action actually reduced the risk it was meant to address, rather than simply being marked complete without verification.
Common Mistakes That Turn FMEA Into a Shelf Document
Scoring Once, Never Updating
FMEA is a living document. When a corrective action is implemented, the affected failure mode should be re-scored to confirm the risk actually decreased — most registers are built once for an audit and never touched again.
Disconnecting Scores From Maintenance Tasks
A high-RPN failure mode that doesn't translate into an actual PM task, inspection interval, or monitoring trigger is analysis without action — the register exists, but nothing on the floor changed because of it.
Inconsistent Scoring Across Team Members
Without an anchored, shared scoring guide, the same failure mode gets rated differently by different engineers, making the resulting priority ranking unreliable and difficult to defend to plant leadership.
Ignoring Action Priority for High-Severity Items
Sorting strictly by RPN and applying a single cutoff threshold can bury a genuine safety-critical failure mode beneath frequent but low-consequence issues, exactly the blind spot Action Priority scoring is designed to catch.
Building FMEA Into an Ongoing Reliability Cadence
A single FMEA cycle, however thorough, is a snapshot. Equipment ages, operating conditions shift, and new failure history accumulates every month — which means a register that was accurate a year ago can drift out of alignment with reality if it's never revisited. A sustainable program treats FMEA as a recurring discipline rather than a one-time project completed ahead of an audit.
Annual Full Review
Once a year, revisit the complete register for the scoped asset or line — checking whether Occurrence scores still match actual failure frequency, whether new failure modes have emerged, and whether previously low-priority items have drifted upward as equipment ages.
Trigger-Based Updates
Outside the annual cycle, update the affected failure modes immediately whenever a corrective action is implemented, a design change is made, or an unexpected failure occurs that the register didn't anticipate — each of these is a signal that a specific score needs re-verification now, not at the next scheduled review.
New Asset Onboarding
Every newly commissioned asset or major equipment upgrade should receive its own FMEA before or shortly after startup, rather than waiting for the next full-plant review cycle to catch it — early failure modes are often the most preventable if identified before they've had a chance to occur even once.
The teams that sustain FMEA successfully tend to share one habit: they treat the register as a working input to maintenance planning meetings, not a document reviewed only when an auditor asks for it. When the top-ranked failure modes from the FMEA register show up directly in the next planning period's PM schedule and inspection routes, the connection between analysis and action stays visible to everyone involved — which is also what keeps the next annual review from becoming a rushed, disconnected exercise all over again.
FMEA is applied at different stages of an asset's life, and confusing which type is appropriate for a given question is a common source of wasted effort. Design FMEA examines a product or component's design before it exists in physical form. Process FMEA examines the manufacturing process steps used to produce something. Asset-level FMEA — the type most relevant to maintenance and reliability teams — examines an already-installed piece of equipment to determine its ongoing maintenance strategy.
DFMEA — Design Stage
Applied before a product or component is manufactured, examining how the design itself could fail — tolerance stack-up, material selection, or a design feature that concentrates stress in a way that invites fatigue failure.
PFMEA — Process Stage
Applied to the manufacturing process that produces a part, examining how a process step could produce a defect — incorrect torque, misaligned fixturing, or a temperature excursion during a heat treatment step.
Asset FMEA — Reliability Stage
Applied to installed equipment already running in production, examining how the asset itself could fail during operation — the type that directly informs PM task frequency, inspection routes, and spare parts stocking decisions.
Stop Letting Your FMEA Register Live in a Spreadsheet
iFactory connects every failure mode's RPN and Action Priority score directly to a maintenance task, an inspection route, or a monitoring trigger — so the analysis your team already did actually changes what happens on the floor.
Frequently Asked Questions
How is asset-level FMEA different from a standard criticality analysis?
Criticality analysis typically ranks whole assets by their overall importance to production — which machine matters most if it goes down. FMEA goes a level deeper, breaking each asset into its individual failure modes and scoring each one separately on Severity, Occurrence, and Detection. A single critical asset might have a dozen distinct failure modes, only a few of which actually carry high risk once properly scored. Visit support to see how the two analyses connect in practice.
How often should an FMEA register be updated?
FMEA should be treated as a living document, updated whenever plant conditions, failure history, or maintenance strategy changes meaningfully — not scored once and filed away. A practical minimum is an annual full review plus updates whenever a corrective action is implemented, so the affected failure mode can be re-scored to confirm the risk actually decreased as intended.
What's a reasonable RPN threshold for triggering a corrective action?
There's no universal number, because RPN thresholds should be calibrated to your own facility's risk tolerance and historical data — but regardless of the numeric threshold chosen, any failure mode with a Severity score of 9 or 10 should require action regardless of its calculated RPN, since the consequence is unacceptable even at low frequency. Book a demo to see how Action Priority scoring handles this automatically.
Can FMEA be conducted without historical failure data?
It can, but the resulting scores will be less reliable — Occurrence and Detection ratings based purely on team intuition tend to drift from actual failure patterns. Where CMMS work order history, OEM failure data, or industry benchmarks exist, cross-checking scores against them substantially improves the accuracy and defensibility of the resulting priority ranking. Contact support for guidance on connecting existing maintenance history to a new FMEA analysis.
Who should be involved in scoring an asset-level FMEA?
A cross-functional team produces materially better results than a single engineer working alone — maintenance technicians who know the actual failure history, operators who understand real operating conditions, and reliability or quality engineers who can anchor scores to a consistent methodology. Scoring done in isolation by one person tends to miss failure modes that only become visible from a different vantage point on the floor.
