The Repeat Incident: Organisational Amnesia and the Failure to Learn

At a chemical processing facility, a reactive chemical spill occurred during a transfer operation. The investigation identified the root cause: the transfer procedure did not include a step for verifying that the receiving vessel was correctly aligned. A recommendation was issued to revise the procedure. The procedure was revised. Two years later, the same type of spill occurred during the same type of transfer. The investigation found the same root cause. The revised procedure existed in the document management system. The operators had never been trained on it. The procedure revision had been treated as a document management task, not an operational change.

The third occurrence, eighteen months after the second, happened after a site leadership change. The new operations manager, unfamiliar with the history, questioned why the transfer procedure was "so complicated" and authorised a simplified version. The simplification removed the verification step that had been added after the first incident. The organisation had not only failed to learn — it had unlearned.

ORGANISATIONAL AMNESIA Organisational amnesia is not forgetting. It is the structural inability to retain and transmit investigation learning across time, leadership changes, and organisational boundaries. The knowledge exists somewhere — in a report, in someone's memory, in a procedure footnote. But it is not accessible to the person making the decision that will determine whether the incident recurs.

Cause 1: Knowledge lives in people, not systems

In most organisations, investigation knowledge resides in the memory of the people who conducted the investigation. When those people move, retire, or are reassigned, the knowledge goes with them. The investigation report exists in a filing system, but it is a static document that is rarely consulted. The critical context — why certain decisions were made, what the investigation nearly missed, which recommendations were most important — is lost with the departure of the investigators.

Chris Argyris distinguished between "espoused theory" (what the organisation says it does) and "theory-in-use" (what it actually does). Investigation reports capture espoused theory. The theory-in-use — how the investigation actually changed practice — lives in operational memory, which is fragile and perishable.

KEY TAKEAWAY If your investigation learning would be lost if the lead investigator left the organisation tomorrow, your knowledge management system has failed. Investigation learning must be embedded in systems, not stored in people.

Cause 2: Leadership turnover erases memory

The average tenure of a site manager in high-hazard industries is two to four years. Every leadership change creates a risk of institutional memory loss. New leaders inherit procedures but not the context behind them. They may simplify, modify, or deprioritise controls that were implemented in response to specific incidents — because they do not know the incident history that justified those controls.

Cause 3: Single-loop learning dominates

Argyris's framework distinguishes between single-loop learning (fixing the problem within existing frameworks) and double-loop learning (questioning and changing the frameworks themselves). Most post-incident responses are single-loop: revise the procedure, retrain the worker, add a checklist. The underlying system design, the assumptions behind it, and the organisational conditions that enabled the failure remain unexamined. Single-loop learning produces corrections. Only double-loop learning produces prevention.

1. Investigation knowledge base: Create a searchable database of all investigation findings, indexed by hazard type, contributing factor, management system element, and GFT. Before any new investigation begins, the team must search the database for previous incidents with similar characteristics. This prevents the reinvestigation of known failures.
2. Incident history briefing for new leaders: Every incoming site manager, operations manager, or HSE director must receive a structured briefing on the site's investigation history: what happened, what changed, and why specific controls exist. This briefing should be a formal element of the leadership transition process, not a courtesy conversation.
3. Procedure provenance tracking: Every safety-critical procedure should carry a provenance note: when it was last revised, why, and which incident or investigation triggered the revision. When a new leader proposes simplification, the provenance note provides the context that prevents unlearning.
4. Annual repeat incident analysis: Conduct an annual analysis of all incidents to identify repeat patterns — same hazard type, same contributing factors, same management system failures. If any pattern repeats more than twice, the response is not another investigation but a management system redesign.
5. Double-loop learning triggers: Define specific triggers that automatically escalate from single-loop to double-loop learning: any repeat incident, any SIF-potential event, any investigation that identifies a management system design failure. When these triggers activate, the response shifts from "fix the procedure" to "redesign the system."

THE AMNESIA TEST Search your incident database for the last five years. Identify any hazard type or contributing factor that appears in three or more investigations. Each repeat is evidence of organisational amnesia. The number of repeats is the measure of your organisation's failure to learn.

"The repeat incident is the most honest audit of your management system. It tells you exactly what the organisation learned from the last investigation: nothing." — Bruno Hounkpati