Noah Steiner·
Traced our on-time-delivery drop through 5 Whys to a cycle-count cut nobody connected to the problem
Deconstruct complex problems using multiple root cause frameworks (5 Whys, Fishbone, Fault Tree) with evidence validation and solution mapping.
Root Cause Analysis & Problem Deconstruction Engine
You are a master problem solver trained in Six Sigma Black Belt, Toyota Production System, and systems thinking. Conduct a rigorous root cause analysis.\n\nPROBLEM STATEMENT: {{problem_description}}\nIMPACT OF PROBLEM: {{quantified_impact — e.g., '$500K annual cost', '23% customer churn increase', '4-hour system downtime monthly'}}\nWHEN IT STARTED: {{timeline — e.g., '6 months ago', 'Gradual over 2 years'}}\nWHAT HAS BEEN TRIED: {{previous_solutions_attempted}}\n\nOUTPUT — Multi-Framework Root Cause Analysis:\n\n## 1. PROBLEM CLARIFICATION\nBefore analyzing causes, ensure we're solving the right problem:\n\n**Problem Statement Refinement** (SMART format):\n- Specific: What exactly is happening? (symptom)\n- Measurable: How often, how much, what magnitude?\n- Affected: Who/what is impacted?\n- Relevant: Why does this matter strategically?\n- Time-bound: When did it start and what's the trend?\n\n**Is this a problem or a symptom?** Ask: If we 'fixed' this, would something worse appear? Trace upward.\n\n**Problem Type Classification**: [Simple / Complicated / Complex / Wicked]\n- Simple: Clear cause-effect, known solution\n- Complicated: Multiple causes, known solutions exist but need diagnosis\n- Complex: Emergent patterns, solutions emerge through experimentation\n- Wicked: No clear definition, no stopping rule, solutions are contested\n\n## 2. 5 WHYS ANALYSIS\nChain of causality — don't stop at symptoms:\n\n**Why 1**: Why is {{problem_description}} happening? → [Answer]\n**Why 2**: Why is [Answer] happening? → [Answer]\n**Why 3**: Why is [Answer] happening? → [Answer]\n**Why 4**: Why is [Answer] happening? → [Answer]\n**Why 5**: Why is [Answer] happening? → [ROOT CAUSE HYPOTHESIS]\n\nGenerate 2-3 separate 5 Whys chains exploring different causal pathways:\n\n**Chain A: [Process/Operational pathway]**\n[5 Whys chain]\n\n**Chain B: [People/Skills pathway]**\n[5 Whys chain]\n\n**Chain C: [System/Structural pathway]**\n[5 Whys chain]\n\n## 3. ISHIKAWA (FISHBONE) DIAGRAM — TEXT VERSION\nOrganize potential causes by category:\n\n**PEOPLE** (human factors):\n- [Cause 1]: Evidence? [ ]\n- [Cause 2]: Evidence? [ ]\n- [Cause 3]: Evidence? [ ]\n\n**PROCESS** (workflow factors):\n- [Causes with evidence checks]\n\n**TECHNOLOGY/TOOLS** (equipment, software, systems):\n- [Causes with evidence checks]\n\n**DATA/INFORMATION** (information flow, accuracy, availability):\n- [Causes with evidence checks]\n\n**ENVIRONMENT** (external conditions, market, regulations):\n- [Causes with evidence checks]\n\n**MANAGEMENT/GOVERNANCE** (decision-making, oversight, incentives):\n- [Causes with evidence checks]\n\n## 4. FAULT TREE ANALYSIS\nFor critical failures, build a fault tree:\n\nTOP EVENT: {{problem_description}}\n\n**AND gates** (all sub-events must occur):\n- [Sub-event A] AND [Sub-event B] → [Intermediate event]\n\n**OR gates** (any sub-event can trigger):\n- [Sub-event A] OR [Sub-event B] → [Intermediate event]\n\n**Basic events** (root causes, no further decomposition needed):\n- [Cause 1] — Probability estimate: [High/Med/Low]\n- [Cause 2] — Probability estimate: [High/Med/Low]\n\n## 5. CAUSE VALIDATION\nDistinguish correlation from causation:\n| Potential Cause | Evidence of Correlation | Evidence of Causation | Temporal Precedence? | Alternative Explanation? | Validated? |\n|-----------------|------------------------|-----------------------|---------------------|-------------------------|------------|\n\nValidation tests:\n- **If [cause], then [effect] should follow**: Test with historical data\n- **If [cause] removed, [effect] should diminish**: Check past interventions\n- **If [cause] present elsewhere, [effect] appears there too**: Cross-sectional check\n\n## 6. CONTRIBUTING FACTORS RANKING\n| Rank | Root Cause | Impact Contribution | Fixability | Evidence Strength | Validation Method | Priority |\n|------|-----------|--------------------|-----------|------------------|-------------------|----------|\n\n## 7. SOLUTION MAPPING\nFor each validated root cause:\n\n| Root Cause | Solution Category | Specific Solution | Effort | Impact | Feasibility | Quick Win? | Dependencies |\n|-----------|-------------------|-------------------|--------|--------|-------------|------------|--------------|\n\nSolution categories:\n- **Eliminate**: Remove the cause entirely (best)\n- **Replace**: Substitute with something better\n- **Simplify**: Reduce complexity where errors occur\n- **Detect**: Add early warning systems\n- **Contain**: Limit damage when cause occurs (last resort)\n\n## 8. CORRECTIVE ACTION PLAN\n- Immediate containment actions (today/this week)\n- Short-term corrective actions (30 days)\n- Long-term preventive actions (90 days)\n- Verification method: How will we know the root cause is truly fixed?\n- Monitoring plan: Leading indicators to watch\n- Feedback loop: When to revisit if problem recurs
Ergebnisse
# Root Cause Analysis — On-Time Delivery Dropped from 94% to 78%
**Problem:** OTD fell 16pts over 4 months. **Impact:** ~€220k/quarter in penalties + churn risk. **Tried:** added a third shift (no improvement).
## 1. Clarification
Symptom = late deliveries. Type: **Complicated** (multiple known-ish causes needing diagnosis). The third shift didn't help → the bottleneck isn't labor capacity.
## 2. 5 Whys (process chain)
- Why late? Orders miss the dispatch cutoff.
- Why missed? Picking finishes late.
- Why? Pickers wait on stock that shows "available" but isn't on the shelf.
- Why? Inventory counts drift from reality.
- Why? **Cycle counts were cut during a cost-saving push** → root cause.
## 3. Fishbone (extract)
- **Process:** cycle-count frequency reduced — evidence: count cadence halved in Q1.
- **Data:** WMS stock != physical — evidence: 6% discrepancy in audit.
- **People:** new pickers undertrained — evidence: error rate 2× for <90-day staff.
## 4. Cause Validation
| Cause | Correlation | Temporal precedence | Validated? |
|-------|-------------|---------------------|------------|
| Cycle-count cut | OTD drop began the same month | Yes | Likely |
| Picker training gap | Errors up but predates drop | Partial | Contributing |
## 5. Solution Mapping
| Root Cause | Solution | Effort | Impact | Quick win |
|------------|----------|--------|--------|-----------|
| Inventory drift | Restore daily ABC cycle counts | Low | High | Yes |
| Training gap | Buddy system for new pickers | Low | Med | Yes |
## 6. Corrective Plan
Now: reinstate cycle counts + lock the dispatch cutoff. 30 days: WMS reconciliation. Verify: OTD back above 90% within 6 weeks; watch stock-discrepancy % as the leading indicator.
Modell: Claude Sonnet 4
7 Likes3 SavesScore: 4