> read /field-guides/fishbone

FIELD GUIDE #002 // POSSIBLE CAUSES

FishboneA diagram of possibilities, not proof.

Fishbone analysis helps you widen the investigation before you narrow it. The branches organize plausible causes so the team can see what it has considered, what it has ignored, and what still needs evidence.

USE WHENMANY CAUSES COULD EXPLAIN THE EFFECT
AVOID WHENYOU ALREADY HAVE A CLEAR SINGLE CHAIN
REMEMBERBRANCHES ARE HYPOTHESES

START WITH THE EFFECT_

Write the observable problem at the head of the fish. Then build cause families around it. Categories are prompts for coverage, not boxes that reality must obey.

CAPYBARA RULE: A cause written on the diagram has earned visibility, not credibility. Evidence decides credibility.

MAP BROADLY, THEN NARROW_

01Define the effect. Make it concrete enough that everyone is investigating the same thing.
02Use cause families. People, Process, Tools / Systems, Environment, Inputs and Measurement are useful prompts — rename them when the problem demands it.
03Add plausible causes. Capture alternatives before arguing about which one is “the root cause.”
04Mark evidence status. Untested, no, likely or confirmed should mean something observable.
05Choose the next test. Ask which observation would most efficiently separate competing explanations.

CAPYBARA CASE #002_

In The Mysterious Muddy Pool, the muddy water could have come from weather, vegetation, animals, maintenance, equipment or water conditions. The diagram keeps those alternatives visible.

Evidence eventually supports a path through missed maintenance → clogged filter → low flow → muddy pool. The important point is not that the diagram “found” the cause. The investigation used evidence to eliminate or strengthen branches until one path deserved deeper attention.

COMMON FAILURE MODES_

Category theatre

Filling every branch because the template has six categories creates noise. Coverage matters; decorative completeness does not.

Voting on the cause

Popularity is not evidence. A team can be unanimously wrong.

Using vague causes

“Communication” or “people” is too broad to test. Rewrite the branch as a condition you could observe.

Stopping at the map

A Fishbone with twenty causes and no next test is a brainstorming artifact, not an investigation.

WHEN TO HAND OFF_

Once one branch becomes worth deeper causal work, send it into 5 Whys. If the effect is recurring and you have category data, Pareto can help decide which effect deserves attention first.

GOOD NEXT-TEST QUESTION: What result would make one plausible cause less likely while making another more likely?

OPEN FISHBONE →USE IN AN INVESTIGATION →