Skip to content
Problem Solving Foundations
Module 3 of 5

Ishikawa: From Cause Ideas to Evidence Tests

8 minutes · Beginner

Learning objective

By the end of this lesson, you can convert a broad fishbone diagram into a prioritised evidence plan.

Why it matters

A full diagram can feel like progress while leaving the team with dozens of untested ideas. The critical transition is from branch to test.

Core concept: prioritise for learning

Score candidate causes on:

  • potential impact;
  • plausibility from current evidence;
  • size of the evidence gap;
  • cost and speed of testing;
  • value of knowing.

Do not rank by how many people mentioned the cause.

Visual explanation

Worked example

Candidate: “The messaging provider delayed reminders.”

Test:

  • compare provider delivery timestamps before and after the complaint increase;
  • compare wrong-content complaints with delivery latency;
  • inspect whether the wrong appointment time existed before the message was sent.

If latency is unchanged and payloads were already wrong, the provider branch is rejected for this effect. It may still matter for another problem.

Common mistake

Mistake: choosing a cause by workshop vote.

Voting reflects belief, authority and availability—not causation. Vote on which hypothesis to test first, not which one is true.

Quick check

Which is the best next step after brainstorming?

A. Fix the branch with the most notes.
B. Select high-value hypotheses and define evidence that could reject them.
C. Ask the most senior participant for the root cause.
D. Publish the diagram.

Answer: B.

Practical prompt

Select three causes from a fishbone diagram. For each, write one observation or comparison that could support it and one result that would reject it.

Summary

  • The diagram organises hypotheses; it does not verify them.
  • Prioritise by learning value and impact.
  • Define tests that can reject a preferred explanation.

Next lesson

Continue to Pareto Analysis to focus on the categories with the largest measured contribution.

Next step

Pareto Analysis: Choose the Right Measure

Continue