fmea

# FMEA (Failure Mode Effects Analysis) ## Overview FMEA is a systematic, proactive method for evaluating processes to identify where and how they might fail and assessing the relative impact of different failures. Developed in aerospace/automotive industries, FMEA calculates Risk Priority Numbers (RPN = Severity × Occurrence × Detection) to prioritize which failures to prevent first. ## When to Use - Designing new products, processes, or systems - Launching critical features with high failure cost - Incident post-mortems to prevent recurrence - Evaluating vendor/supplier reliability - Improving manufacturing quality - Pre-launch risk assessment ## The Process ### Step 1: Identify Potential Failure Modes List all ways the system could fail. For each component/process step, ask: "What could go wrong?" **Example:** E-commerce checkout process failure modes: payment gateway timeout, inventory out of sync, address validation fails, credit card decline, email confirmation fails. ### Step 2: Assess Severity (1-10 Scale) Rate impact if failure occurs. 1 = no impact, 10 = catastrophic. Focus on customer/business impact, not technical complexity. **Example:** Payment gateway timeout: Severity = 9 (lost revenue, customer frustration). Email confirmation fails: Severity = 3 (inconvenient but order proceeds). ### Step 3: Assess Occurrence (1-10 Scale) Rate likelihood of failure happening. 1 = rare, 10 = inevitable. Base on data if available, educated guess if not. **Example:** Payment timeout: Occurrence = 2 (gateway 99.5% uptime). Email fails: Occurrence = 4 (email service less reliable). ### Step 4: Assess Detection (1-10 Scale) Rate likelihood of catching failure before customer impact. 1 = always detected, 10 = never detected until customer complains. **Example:** Payment timeout: Detection = 3 (monitoring alerts immediately). Email fails: Detection = 8 (no monitoring, customer must report). ### Step 5: Calculate RPN and Prioritize RPN = Severity × Occurrence × Detection. Higher RPN = higher priority. Focus mitigations on highest RPNs first. **Example:** - Payment timeout: 9 × 2 × 3 = 54 RPN - Email fails: 3 × 4 × 8 = 96 RPN → Prioritize email monitoring despite lower severity ### Step 6: Implement Actions and Reassess Design mitigations to reduce Severity (safer failure mode), Occurrence (prevent failure), or Detection (catch earlier). Recalculate RPN after mitigation. **Example:** Add email delivery monitoring (Detection 8 → 2). New RPN: 3 × 4 × 2 = 24 (acceptable). ## Example Application **Situation:** SaaS company launching new API with high-value enterprise customers. **Failure Modes Analysis**: 1. API returns 500 errors: S=10, O=3, D=2 → RPN=60 2. Response time >5sec: S=7, O=5, D=4 → RPN=140 3. Authentication fails: S=9, O=2, D=3 → RPN=54 4. Rate limiting blocks requests: S=8, O=6, D=7 → RPN=336 **Prioritization**: Address rate limiting first (RPN=336), then response time (RPN=140). **Mitigations**: - Rate limiting: Per-customer limits (O: 6→2), monitoring alerts (D: 7→2) → New RPN=32 - Response time: Database indexing (O: 5→2), performance monitoring (D: 4→1) → New RPN=14 **Outcome:** Launch succeeds, no customer-facing failures. Prevented potential churn. ## Anti-Patterns - ❌ Skipping low-severity, high-occurrence issues (death by a thousand cuts) - ❌ Rating based on technical complexity vs. customer impact - ❌ Ignoring detection rating (undetected failures multiply impact) - ❌ Treating FMEA as one-time exercise vs. living document - ❌ Analysis paralysis (don't FMEA every trivial process) - ❌ Confusing RPN with absolute risk (RPN is relative prioritization tool) ## Related - five-whys - fishbone-diagram - root-cause-analysis - risk-management - pre-mortem