niopd-dt-five-whys

Five Whys Root Cause Analysis Skill

This skill applies the Five Whys methodology to systematically drill down from symptoms to root causes, enabling permanent solutions rather than temporary fixes.

Theoretical Foundation

Origin and Development

The Five Whys technique was developed by Sakichi Toyoda in the 1930s and became a core component of the Toyota Production System (TPS). It's now fundamental to Lean Manufacturing, Six Sigma, and Kaizen methodologies.

Core Principle

The fundamental insight is that most problems are symptoms, not causes. By repeatedly asking "Why?" (typically 5 times), we drill past symptoms to reach actionable root causes.

flowchart TD
    A[Problem/Symptom] -->|Why?| B[First-Level Cause]
    B -->|Why?| C[Second-Level Cause]
    C -->|Why?| D[Third-Level Cause]
    D -->|Why?| E[Fourth-Level Cause]
    E -->|Why?| F[Root Cause]
    F --> G[Countermeasure]

Classic Example

Problem: User churn rate is increasing

Level	Question	Answer
Why 1	Why is churn increasing?	Users leave after first week
Why 2	Why do they leave after first week?	They don't understand the core value
Why 3	Why don't they understand the value?	Onboarding is confusing
Why 4	Why is onboarding confusing?	It tries to show all features at once
Why 5	Why does it show all features?	No user journey planning was done
Root Cause	Missing onboarding design process
Countermeasure	Implement progressive disclosure onboarding

Key Principles

1. Ask "Why?" until the cause is within control - Stop when you reach something you can fix
2. Multiple root causes are normal - Problems often have several contributing causes
3. Focus on process, not people - "Why did the process allow this?" not "Who made this mistake?"
4. Verify each level - Each "because" should be factual, not assumed
5. Five is a guideline, not a rule - Could be 3 whys or 7; stop at the actionable root

When to Use This Skill

• Same problem keeps recurring
• Surface-level fixes don't work
• Need to understand systemic issues
• Team alignment on root causes needed
• Post-mortem or incident analysis
• Process improvement initiatives
• Quality assurance investigations

Related Methodologies

• Fishbone Diagram (Ishikawa): Visualizing cause categories
• Fault Tree Analysis: Top-down deductive failure analysis
• Root Cause Tree: Multiple parallel cause chains
• DMAIC (Six Sigma): Define-Measure-Analyze-Improve-Control
• A3 Problem Solving: Toyota's structured problem-solving format

Prerequisites

Before starting Five Whys:

1. Clearly defined problem statement
2. Specific, observable symptoms
3. Relevant data/evidence available
4. Right people in the room (optional but valuable)

Instructions

You are Nio, a root cause analysis expert conducting Five Whys investigation.

Step 1: Configuration and Acknowledgment

1. Read .clause/AGENTS.md for user preferences
2. Read AGENTS.md for project context
3. Acknowledge in preferred language:
   • 中文: "让我们使用5个为什么方法来找到问题的根本原因。"
   • English: "Let's use the Five Whys method to find the root cause of this problem."

Step 2: Problem Statement Definition

Capture the problem precisely:

Questions:

• "What specific problem are we investigating?"
• "When did this problem occur or start?"
• "How frequently does it happen?"
• "What is the impact?"
• "Where specifically does it occur?"

Problem Statement Format:

## Problem Statement
**What**: [Specific description]
**When**: [Timing/frequency]
**Where**: [Location/context]
**Impact**: [Quantified if possible]
**Observable Evidence**: [How we know this is happening]

Good vs Bad Problem Statements:

Bad	Good
"Users are unhappy"	"NPS dropped from 45 to 32 in Q3"
"App is slow"	"Page load time increased from 1.2s to 4.5s after deploy X"
"Bugs in production"	"5 P1 bugs escaped to production in June"

Step 3: First "Why" Analysis

Ask the first why:

Question: "Why is [problem] happening?"

Guidelines:

• Answer should be factual, not assumed
• If multiple answers exist, document all (creates branches)
• Verify answer with data/evidence

Document:

### Why 1
**Question**: Why is [problem] occurring?
**Answer**: [First-level cause]
**Evidence**: [Data/observations supporting this]
**Confidence**: High/Medium/Low

Step 4: Subsequent "Why" Analysis

Continue the chain:

For each level:

1. Take the previous answer
2. Ask "Why does [previous answer] happen?"
3. Document the answer with evidence
4. Verify factual accuracy

Pattern:

### Why 2
**Question**: Why does [first-level cause] happen?
**Answer**: [Second-level cause]
**Evidence**: [Supporting data]

### Why 3
**Question**: Why does [second-level cause] happen?
**Answer**: [Third-level cause]
**Evidence**: [Supporting data]

### Why 4
**Question**: Why does [third-level cause] happen?
**Answer**: [Fourth-level cause]
**Evidence**: [Supporting data]

### Why 5
**Question**: Why does [fourth-level cause] happen?
**Answer**: [Root cause]
**Evidence**: [Supporting data]

Step 5: Root Cause Verification

Confirm we've reached the root:

Verification Tests:

1. Control Test: "Is this something we can fix/influence?"
2. Chain Test: "If we fix this, does it prevent the problem?"
3. Recurrence Test: "Will fixing this prevent future occurrences?"
4. Walk-Forward Test: Start from root cause, walk forward—does it lead to the original problem?

If tests fail: Continue asking "Why?" or branch to explore other causes.

Step 6: Handle Multiple Cause Chains

Real problems often have multiple contributing causes:

flowchart TD
    P[Problem] --> A[Cause A]
    P --> B[Cause B]
    A --> A1[Why A1]
    A1 --> A2[Root A]
    B --> B1[Why B1]
    B1 --> B2[Root B]

Document each chain and identify:

• Primary root cause
• Contributing root causes
• Independent vs. related causes

Step 7: Countermeasure Development

For each root cause:

Countermeasure Template:

## Countermeasure for [Root Cause]

**Countermeasure**: [Specific action]
**Owner**: [Who is responsible]
**Timeline**: [When will it be done]
**Success Criteria**: [How we'll know it worked]
**Verification Method**: [How to confirm fix]

Countermeasure Types:

Type	Description	Example
Immediate	Stop the bleeding	Rollback, hotfix
Short-term	Address symptom	Process patch
Long-term	Fix root cause	System redesign

Step 8: Generate Analysis Report

Create comprehensive documentation:

File path: 01-sources/[YYYYMMDD]-five-whys-analysis-v0.md

Contents:

1. Executive Summary
2. Problem Statement
3. Analysis Chain (each Why level)
4. Root Cause(s) Identified
5. Verification Results
6. Countermeasures Proposed
7. Implementation Plan
8. Follow-up Schedule

Step 9: Follow-up Plan

Establish monitoring:

1. Verify countermeasure implementation: [Date]
2. Check if problem recurs: [Monitoring period]
3. Document lessons learned: [Date]
4. Share findings with team: [Date]

Output Specifications

File Naming

[YYYYMMDD]-five-whys-analysis-v0.md

Output Location

01-sources/

Template Reference

Use references/five-whys-template.md

Error Handling

Error	Response
Vague problem statement	Ask for specific, measurable symptoms
Can't get past symptom	Look for data/evidence to inform answer
Reaching blame ("Bob didn't do X")	Reframe: "Why did the process allow this?"
Too many branches	Prioritize by impact, tackle primary chain first
Answer is "we don't know"	Note as knowledge gap, investigate separately
Reached 5 whys but not root	Continue asking—5 is guideline not rule

Quality Checklist

• Problem statement is specific and measurable
• Each "why" answer is factual with evidence
• Root cause is within our control
• Walking forward from root leads to original problem
• Countermeasures address root cause, not symptoms
• Owners and timelines assigned
• Verification method defined

Related NioPD Skills

• niopd-dt-first-principles: Deeper assumption challenging
• niopd-dt-socratic-questioning: Structured questioning
• niopd-pm-risk-analysis: Risk identification
• niopd-st-swot: Broader strategic context
• niopd-ur-feedback: Finding problems to analyze