QM Bench

Fishbone (Ishikawa) diagram: theory, step-by-step guide and case study

How to run a cause and effect analysis with the 6M categories, how to go from a long list of possible causes to verified root causes, and a worked example from machining.

Ľubomír Blažek
Automotive quality manager · updated 2026-10-25

Try it while you read

Open the free Fishbone Diagram tool and click Load sample: it reproduces the case study below.

Open the tool

Theory: structured brainstorming

The cause and effect diagram, developed by Kaoru Ishikawa, organises the possible causes of a problem into categories so that the team does not jump to the first idea. The problem (effect) sits at the head of the “fish”, the categories form the main bones, and causes and sub-causes branch off them.

The classic 6M categories for manufacturing are Man (people, skills), Machine (equipment, tools), Method (procedures, settings), Material (raw material, components), Measurement (gauges, inspection) and Environment (temperature, dust, vibration). They act as a checklist: if a category is empty, the team has probably not thought about it.

From possible to verified

A fishbone lists hypotheses. The value comes from the next step: choosing the most likely causes and proving them with data, trials or a “switch on / switch off” test. Only verified causes justify corrective actions.

Key point

Combine the fishbone with 5 Why: the diagram broadens the search, 5 Why goes deep on the most likely branch.

Step by step

  1. Define the effect precisely: “Shaft diameter out of tolerance on line 3”, not “quality problem”.
  2. Gather the team: operator, setter, quality, maintenance.
  3. Fill the categories, one cause per line. Start a line with “-” for a sub-cause.
  4. Ask why until a cause is specific and actionable.
  5. Vote on the most likely causes, then plan how to verify each.
  6. Mark verified causes with “*” and add the diagram to the 8D (D4) with one click.

Case study: shaft diameter out of tolerance

After a customer complaint about oversize bearing seats, the team fills a fishbone (the Load sample data): 16 possible causes across the 6 categories, for example “new operator not trained”, “spindle thermal growth, no warm-up cycle”, “micrometer not calibrated” and “no tool-wear compensation, offset not in instruction”.

The three most likely causes are tested: a warm-up trial shows only 3 µm thermal growth (not the cause); the micrometer is within calibration (not the cause); measuring parts before and after an insert change shows a jump of 0.02 mm because the offset is not corrected. “No tool-wear compensation” is marked as verified.

Result

The verified cause goes into D4 of the 8D report. The action, automatic probing after every insert change, addresses it directly.

Common mistakes

  • Vague causes (“human error”, “bad material”).
  • Treating every listed cause as true and acting on all of them.
  • Doing the diagram alone at the desk instead of with the people at the process.

Frequently asked questions

Do I have to use the 6M categories?

No. 6M works well for production. For services or office processes, categories like People, Process, Policies, Place and Technology are common. Rename or add categories as needed.

How many causes should a fishbone have?

Quality over quantity: 15 to 30 specific causes are typical. A cause like "operator error" is too vague; "offset entered wrongly after insert change" is actionable.