Overview
The Ishikawa (or fishbone, or cause-and-effect) diagram puts the problem at the head of a fish and branches the possible causes off the spine, grouped into categories. It forces a team past the first cause that comes to mind and onto the full landscape of contributing factors before anyone rushes to a fix.
Its strength is breadth. By pre-seeding categories — classically People, Process, Equipment, Materials, Environment, Management — it prompts you to look in places you would otherwise skip, surfacing causes that a single-thread investigation misses.
When to use it
A recurring problem with no single obvious cause — defects, outages, dropping metrics.
How to use it
State the problem
Write the specific effect at the head of the fish.
Draw the category bones
Add a branch for each cause category relevant to your domain.
Brainstorm causes per category
Under each branch, ask what could contribute. Add sub-branches for causes of causes.
Ask ‘why’ down each bone
Push each cause deeper, borrowing the 5 Whys, until you reach something actionable.
Identify the likely roots
Circle the causes best supported by evidence, and test those first.
Worked example
A factory’s defect rate spikes. The fishbone spreads causes across People (new hires undertrained), Process (a skipped inspection step), Equipment (a worn tool), and Materials (a new supplier). Investigating the most-supported branches, the worn tool and the skipped step turn out to compound. A single-cause hunt would have blamed the new hires and missed both.
Common pitfalls
- Stopping at symptoms instead of pushing each bone to a root cause.
- Listing causes with no evidence and treating the diagram as proof.
- Using generic categories that do not fit your actual process.
Frequently asked questions
What categories should I use?
The classic 6M (People, Methods, Machines, Materials, Measurement, Environment) is a start — adapt them to your domain.
How is it different from 5 Whys?
5 Whys drills one causal chain deep; the fishbone maps many chains wide. They pair well.
Does the diagram prove the cause?
No — it generates candidate causes. You still have to test which ones are real.