fishbone-analysis

Fishbone (Ishikawa) Analysis

When to use

Use fishbone analysis to brainstorm all possible causes of a quality problem before running 5-Why. It prevents tunnel vision and ensures no cause category is overlooked. Particularly valuable for complex defects with multiple potential contributing factors.

Typical use: 8D D4 brainstorming session, CAPA root cause investigation, initial problem analysis.

Prerequisites

• Problem clearly defined with Is/Is-Not or 5W2H
• Cross-functional team (quality, production, engineering at minimum)
• Access to the process, machine, or product where defect occurred

Required Fishbone Checklist

☐ Problem statement defined and agreed before starting — no cause language in the problem statement ☐ All 6M categories addressed — at least one entry per M, or documented justification if a category is not applicable ☐ Brainstorming completed before any evaluation or elimination — do not evaluate while generating ☐ Every cause classified as Confirmed, Probable, or Unlikely using objective evidence — not opinion alone ☐ Confirmed = supported by data or physical evidence; Probable = logical, consistent with Is/Is-Not, not yet confirmed; Unlikely = contradicted by data ☐ Confirmed and Probable causes cross-checked against existing PFMEA failure cause entries before proceeding to 5-Why ☐ Each Confirmed or Probable cause carries forward to its own 5-Why chain ☐ After root cause confirmed: horizontal deployment check — could the same cause exist in similar parts, processes, or product families?

The 6M Framework

The six main "bones" of the fish. All six must be addressed. If a category genuinely does not apply, document: "No causes identified in this category after structured team review — not applicable because [reason]."

1. Man (Human Factors)

Questions to ask:

• Was the task performed by a trained, qualified operator?
• Is competence documented and current?
• Was the operator following the correct work instruction?
• Could fatigue, shift change, or distraction contribute?
• Is the task ergonomically difficult or error-prone?
• Is the same defect found on all shifts or only one? (single-shift → operator-specific)