Fishbone Diagram: 7 Investigation Traps That Hide Root Causes

A Fishbone Diagram can make incident investigation sharper, or it can make weak analysis look organized. The difference is not the drawing. The difference is whether every branch is tested against evidence, controls, work design, and the decisions that shaped the event before the injured worker appeared in the story.

This article is written for EHS managers, supervisors, and investigation leaders who already know the basic Ishikawa shape but need a stronger way to use it after safety incidents, near misses, and SIF potential events.

Why Fishbone fails when the team starts with opinions

The Fishbone Diagram was designed to organize possible causes, not to certify guesses. In safety, the trap appears when a team enters the room, draws branches on a whiteboard, and starts filling them with familiar explanations before the evidence has been preserved.

James Reason's work on latent failures is useful here because the visible error usually sits near the end of a much longer chain. When the diagram starts with "operator did not follow procedure" and stops there, it misses the procedure nobody could use under production pressure, the supervisor who inherited a weak plan, the maintenance delay that made the workaround normal, and the leadership signal that speed mattered more than escalation.

As Andreza Araujo argues in Sorte ou Capacidade, glossed for English readers as Luck or Capability, accidents should not be treated as bad luck or isolated personal failure. The investigation has to ask what the organization had already made likely.

1. The team fills the branches before preserving evidence

The first trap is timing. A Fishbone session held before photographs, physical evidence, documents, system data, and witness statements are secured can turn the loudest memory into the official cause.

Before the diagram, the team needs an event timeline, preserved scene evidence, equipment status, permit records, training records, maintenance history, supervision notes, and initial witness accounts. Without that foundation, the branches become a catalog of beliefs. They may sound reasonable, although the team has no way to know whether they are true.

Use the principles in incident evidence preservation before scheduling the Fishbone session. The first hour decides whether the analysis will have facts to test or only memories to debate.

2. The categories are too generic for the event

Traditional Fishbone categories such as people, method, machine, material, measurement, and environment can help a team begin. They often fail in serious safety events because they hide the management system inside broad words.

A stronger incident Fishbone uses categories that match how work is actually controlled. For a SIF potential event, the branches may need task planning, supervision, critical controls, competence, equipment integrity, permit-to-work, contractor interface, procurement, maintenance backlog, work pressure, procedure usability, and emergency response.

Across 25+ years leading EHS at multinationals, Andreza Araujo has seen that generic categories make leaders comfortable because they keep causes abstract. Specific categories make ownership harder to avoid. "Machine" is vague. "Guard bypass accepted during restart because maintenance had no spare part" is a management fact.

3. Behavior becomes the final cause instead of a signal

Most weak Fishbone diagrams eventually find a behavior that looks wrong. The worker did not lock out. The technician stood in the line of fire. The driver used the wrong route. The contractor skipped a step. Those facts may be real, but they are rarely sufficient.

The better question is why the behavior made sense in that operating context. Did the task design make the safe step slower than the shortcut? Did the supervisor reward recovery speed? Did the procedure conflict with the real equipment? Did the worker have authority to stop? Did previous deviations pass without correction?

Andreza Araujo's Portuguese title A Ilusao da Conformidade, often explained as The Illusion of Compliance, fits this point. A procedure can exist, training can be recorded, and signatures can be complete while the field system quietly teaches another way to get the work done.

That is why Fishbone should connect with RCA that avoids the operator error trap. The behavior branch is a starting point for analysis, not a place to close the investigation.

4. The diagram ignores control failure

A Fishbone Diagram that does not ask which controls failed is incomplete. The organization needs to know whether the hazard was uncontrolled, whether the control existed but was weak, whether the control was bypassed, or whether nobody verified it before exposure.

This matters most for SIF potential events. A lost-time injury and a fatality can share the same weak barrier, separated only by energy, position, timing, or luck. Heinrich and Bird's accident-pyramid work remains useful because precursor events deserve attention before severity makes the lesson undeniable.

For each branch, ask what control should have prevented the exposure and what evidence proves that control worked. If the answer is a training record, a sign, or a procedure alone, the branch probably needs deeper testing.

Compare the finding with Five Whys for SIFs, because both methods fail when they accept a paper control as if it were a field barrier.

5. Witness statements are used to confirm a theory

Witness statements can enrich a Fishbone Diagram, but they can also be misused. If the investigator already believes the cause, interviews become confirmation work rather than evidence collection.

Daniel Kahneman's work on cognitive bias helps explain the risk. Once a team forms an early theory, it tends to notice evidence that supports the theory and discount details that disturb it. In incident investigation, that bias can turn a diagram into a neat story that protects the organization from harder questions.

Ask witnesses what they saw, heard, expected, and understood at the time. Separate direct observation from interpretation. Then place witness evidence on the branch it actually supports, even when it complicates the first theory.

The interview discipline described in witness statement errors after incidents should happen before the final Fishbone review, not after the action plan has already been decided.

6. The action plan is written from branch labels

A branch label is not an action. If the Fishbone says "training," the action cannot simply be "retrain employees." The team must state which knowledge or skill gap was proven, why the previous training failed, how the new design will be different, and how supervisors will verify the behavior under real work conditions.

The same rule applies to equipment, procedure, supervision, and communication findings. Each action should change a condition that contributed to the event. If the action only reminds people to be careful, the investigation has produced paperwork rather than risk reduction.

During her PepsiCo South America tenure, where the accident ratio fell 50% in six months, Andreza Araujo's work showed that measurable safety improvement depends on changing operating routines, not only issuing new messages after incidents.

Use post-incident action plan controls to translate each Fishbone branch into ownership, deadline, verification, and effectiveness review.

7. The organization never tests whether the root cause changed

The final trap appears after the report is approved. The diagram is stored, actions are closed, and nobody returns to the field to see whether the underlying condition has changed.

In more than 250 cultural-transformation projects supported by Andreza Araujo's team, one recurring pattern is clear: weak organizations close actions by evidence of completion, while stronger organizations close actions by evidence of changed risk. A new procedure, repaired guard, revised permit, or supervisor briefing is only an input until the exposed work proves the control now holds.

Set a verification date while writing the action plan. Observe the task, interview the crew, review leading indicators, and check whether the same branch would still appear if the event happened again. If the answer is yes, the action is administratively closed and operationally open.

Fishbone quality test for EHS managers

Conclusion

A Fishbone Diagram is useful when it slows the team down enough to test evidence, expose latent failures, and translate causes into field changes. It is dangerous when it gives a weak investigation the appearance of structure.

If your organization wants incident investigations that move beyond blame, retraining, and cosmetic closure, Andreza Araujo's ACS Global Ventures consulting work can support a safety culture diagnostic and a stronger investigation-to-action process. Start at Andreza Araujo. Safety is about coming home, including after the investigation teaches the organization what must change.