Fishbone Diagram
Fishbone Diagram: Draw a fish skeleton with the problem as the "head" and major cause categories as "bones." Brainstorm specific causes under each category before investigating. The structure prevents premature fixation on one cause and ensures all contributing factors are considered β essential for complex quality or operational problems with multiple possible explanations.
What Is the Fishbone Diagram?β
The Fishbone Diagram was developed by Kaoru Ishikawa, a Japanese quality management pioneer, in the 1940sβ60s while working on quality improvement at Kawasaki. It became a cornerstone of Total Quality Management and Six Sigma, and remains one of the most widely taught quality tools globally.
The diagram is named for its visual shape: a horizontal arrow pointing to the problem effect (the "head"), with angled branches ("bones") representing categories of potential causes. Smaller sub-branches represent specific causes within each category. The overall structure resembles a fish skeleton.
The primary purpose is comprehensive brainstorming before investigation. Before you spend time testing hypotheses, the fishbone ensures you've generated all plausible hypotheses. The classic error in problem-solving is investigating the first plausible cause, finding it contributes to the problem, implementing a fix, and then discovering the real root cause was elsewhere. The fishbone forces the team to surface all candidates before any investigation begins.
The most common categorisation for manufacturing uses the "6Ms": Machine, Method, Material, Measurement, Mother Nature (environment), and Man (people). Service industries often use "4Ps" (People, Process, Place, Product) or "PEST" (People, Environment, Systems, Technology). The specific categories matter less than ensuring the full causal space is covered.
How It Worksβ
Step 1: Define the problem effect precisely
β Write it at the right side (fish head)
β "Customer complaints about delivery delays" not "delivery problems"
Step 2: Draw the spine and major bones
β Horizontal arrow to the effect
β 4β6 major category bones angled in
Step 3: Label the major bones (use 6Ms or relevant framework)
β Machine, Method, Material, Measurement, Environment, People
β Or customise for your context
Step 4: Brainstorm causes for each bone
β Team brainstorming: what could cause this under each category?
β Add as sub-branches; go 2β3 levels deep
Step 5: Identify most likely causes
β Circle or highlight candidates for investigation
β Prioritise by likelihood and testability
Step 6: Investigate and validate
β Test prioritised hypotheses with data
β Don't stop at one confirmed cause β check others
Three Real-World Examplesβ
Hospital Medication Errorsβ
A hospital experienced elevated medication errors. The fishbone categories:
- People: Nurse fatigue (shift length), inadequate training on new drug protocols, illegible handwriting in paper orders
- Process: No double-check protocol for high-risk medications, workflow interruptions during dispensing
- Equipment: Identical-looking packaging for different drugs, poor labelling on bins
- Environment: Poor lighting in dispensing area, noise levels on ward
- Measurement: No near-miss reporting system, no feedback loop on errors
Without the fishbone, the investigation might have focused on nurse training alone. The fishbone revealed that packaging similarity and the absence of a double-check protocol were equally high-risk β and more fixable than fatigue.
SaaS Product Churn Spikeβ
A SaaS product saw monthly churn rise from 3.5% to 6.2% over Q3.
- Product: New UI redesign launched July 1 (timing correlation), performance regression in v3.2
- Process: Onboarding sequence not updated for new enterprise tier, no proactive health-check outreach
- People: Two senior CSMs left in June β coverage gap for 30 accounts
- Market/External: Competitor launched aggressive pricing campaign in July
- Data/Measurement: Churn attribution model changed in Q3 β could be measurement artefact
Investigation confirmed: the UI redesign confused power users (product), CSM coverage gap meant 12 at-risk accounts received no outreach (process), and competitive pricing pressure accelerated churn in price-sensitive segments (market). All three needed separate interventions.
Manufacturing Defect Rate Increaseβ
A precision parts manufacturer saw defect rate increase from 0.8% to 2.3%.
- Machine: Calibration drift on CNC machine #4, worn tooling not flagged
- Material: New material supplier with slightly different hardness spec
- Method: New operator using non-standard fixturing technique
- Measurement: Quality inspection interval increased from 100 to 500 units last month
- Environment: Summer temperature affecting material expansion tolerances
Root causes: CNC machine #4 calibration (confirmed by measurement) and new material spec (confirmed by test). Method and measurement were contributing factors.
When to Use Itβ
β Fishbone Diagrams are ideal for:
- Quality problems with multiple plausible causes
- Team problem-solving sessions where diverse perspectives are needed
- Situations where the first investigation hasn't found the cause
- Any problem that has likely recurred because prior fixes didn't address the real cause
β Less useful when:
- The cause is already known (don't over-structure obvious problems)
- The problem is too complex for a single diagram (use issue trees instead)
- You need to show causal relationships between causes (use a systems map)
| Pairs well with | Why |
|---|---|
| 5 Whys | Fishbone identifies which causal thread to drill; 5 Whys drills it |
| Root Cause Analysis | Fishbone is one tool within the broader RCA toolkit |
| MECE | Bone categories should be MECE β covering all causes without overlap |
| Pre-mortem | Pre-mortem anticipates fishbone-type causes before problems occur |
Common Misuses and Limitationsβ
Stopping at the fishbone. The diagram is a brainstorming tool, not a diagnosis. Generating causes is Step 1; testing them is Step 2. Many teams build beautiful fishbones and then fail to rigorously validate which causes are actually contributing.
Using generic categories that don't fit the problem. "6Ms" works well for manufacturing; it fits poorly for software or service problems. Adapt the categories to the domain β the goal is comprehensive coverage, not adherence to a specific framework.
Too much depth too early. Three levels of sub-branches on every bone produces an unnavigable diagram. Focus depth on the bones most likely to contain the cause.
Treating it as an individual tool. Fishbone diagrams are most valuable in group settings, where diverse expertise surfaces causes a single analyst would miss. The tool's value scales with the diversity of perspectives in the room.
Related Modelsβ
| Model | Relationship |
|---|---|
| 5 Whys | 5 Whys drills one causal chain deep; Fishbone maps the full causal space breadth-first |
| Root Cause Analysis | Fishbone is one technique within the RCA family |
| Issue Tree | Issue Trees are the structured analysis version; Fishbones are the visual brainstorming version |
| Black Box Thinking | Black Box Thinking is the cultural mindset; Fishbone is a practical implementation tool |
Frequently Asked Questionsβ
Should you use 4Ms, 6Ms, or 8Ms?
Use whatever covers the causal space for your specific problem. Classic manufacturing: 6Ms (Machine, Method, Material, Measurement, Mother Nature, Man). Healthcare often adds "Management" and "Money" for 8Ms. Services often use PESHM (People, Environment, Systems, Hardware, Management). The principle is exhaustiveness β your category set should make it hard for a cause to hide. If a category feels forced or irrelevant, drop it. If there's a major causal domain not covered, add a category.
How do you prioritise which causes to investigate?
After building the diagram, use a combination of: (1) likelihood β which causes have the most plausible causal mechanism given what you know?; (2) timing β which causes align with when the problem started?; (3) testability β which causes can be validated quickly with available data?; (4) impact β if confirmed, which cause would account for most of the problem? Circle 3β5 high-priority causes and investigate those first before expanding the search.
Can the Fishbone Diagram show relationships between causes?
Not natively β the fishbone shows causes leading to an effect, but doesn't show how causes relate to each other. If you need to model cause-cause relationships (this factor amplifies that one), use a systems influence diagram or causal loop diagram instead. For most operational problems, cause-cause interactions aren't essential to model; for complex systems problems, they matter a great deal.
Further Readingβ
- Ishikawa, K. (1986). Guide to Quality Control β the original source
- Montgomery, D.C. (2012). Introduction to Statistical Quality Control β rigorous treatment of quality tools including Fishbone
- Pyzdek, T. & Keller, P. (2014). The Six Sigma Handbook β Fishbone in the DMAIC problem-solving framework
Apply with AIβ
π Build a fishbone diagram for your problem with MindMax β
This page is part of the MindMax Mental Models Knowledge Base.