FMEA
Failure Mode and Effects Analysis
FMEA systematically identifies what CAN go wrong in a product or process — before it happens — and prioritises prevention by severity, occurrence and detectability. It is quality work’s shift from hindsight to foresight.
Use the method when
- →New products, processes or significant changes — BEFORE commissioning
- →Recurring errors show that risks aren’t systematically understood
- →Customer or standard requirements (IATF, ISO 9001 risk thinking) require documented risk analysis
Don’t use it when
- →As a one-off document for an audit — an FMEA that isn’t maintained is outdated at the first change
- →On a whole product/factory in one analysis — bound it to a subsystem or process line, otherwise the seriousness drowns in the volume
Before you start
A cross-functional team: design/process, quality, production/operators — an FMEA made by one person is an opinion list
A process flow diagram or product structure is ready as the backbone
The rating scales (1-10 for severity, occurrence, detection) are agreed in advance — preferably per AIAG-VDA
How to do it — step by step
Bound and structure
1 meetingDefine the analysis’s object and boundaries from the flow diagram: which process steps, which functions. Each step is described with its function/requirement — for a failure is by definition a non-fulfilment of a requirement; unclear requirements give unclear failure modes.
Identify failure modes, effects and causes
1-2 meetingsPer function: how can it fail (failure mode)? What is the consequence for the next link and the end customer (effect)? What can cause it (cause)? Be concrete — “wrong torque on bolt B4” can be analysed; “assembly error” cannot.
Rate S, O and D — disciplined
1 meetingSeverity (S) follows the effect, Occurrence (O) the cause given current prevention, Detection (D) the current control’s ability to catch the error before the customer. Use the agreed scales mechanically — the discussion should be about the facts behind the score, not about negotiating the number down.
Prioritise by action priority
part of a meetingAIAG-VDA’s action priority (AP: high/medium/low) weights severity more heavily than the classic RPN product — an S9 error with a low frequency must never drown under a frequent trifle. High AP requires action; medium requires a justification for any non-action.
Define actions — prevent before detect
1 meeting + follow-upFor each prioritised risk: first preventive actions (design/process change, poka-yoke) that lower the occurrence; then detective ones (control, sensors) that improve detection. Each action has an owner and a deadline, and the score is re-assessed after implementation.
Keep the FMEA alive
ongoingThe FMEA is updated on: changes to product/process, new errors from complaints/CAPA (was the error foreseen? why not?), and a fixed annual review. The link to the control plan is maintained — the FMEA’s detection actions ARE the control plan’s content.
The classic mistakes
How it goes wrong in practice, and this is where most implementations part ways.
FMEA after commissioning
The analysis is done in hindsight to satisfy an audit, and all prevention value is gone. FMEA belongs BEFORE the error can even arise.
Score negotiation
The team negotiates O and D down until everything lands below the action threshold — the analysis becomes a reassuring document. The facilitator’s most important task is to keep the scales honest.
Decoupled from reality
Complaints and internal errors are never confronted with the FMEA. Each unforeseen error is a free lesson in where the analysis was too thin — if it isn’t used, the holes repeat in the next FMEA.
Related standards
The standard is free. The anchoring is the craft.
The method only works when it becomes daily routines and leadership behaviour. The operations check shows in 4 minutes where your operations stand, and what should come first.