The Postmortem Structure

The postmortem is the structured process that transforms a failure event into institutional knowledge. Without a structured postmortem, the failure produces two outcomes: a period of organizational discomfort and a collective determination to do better next time without a specific understanding of what 'better' requires. With a structured postmortem, the failure produces specific findings, specific corrective actions, specific owners, and specific deadlines — and the institutional knowledge that makes the next AI initiative less likely to fail in the same way.

AI INITIATIVE POSTMORTEM
========================
Initiative Name:
Postmortem Date:
Postmortem Lead:
Participants:

SECTION 1: WHAT HAPPENED — FACTUAL TIMELINE
--------------------------------------------
[Chronological account of the initiative from launch to failure event.
Include: key decisions, stakeholder engagements, adoption milestones,
technical events, and the specific failure event. Factual only — no
analysis or attribution in this section.]

Date       | Event
-----------|-------
[Date]     | [Factual description of event]
...

SECTION 2: WHY IT HAPPENED — ROOT CAUSE ANALYSIS
-------------------------------------------------
Primary Root Cause:
[The single most significant factor that produced the failure.
If multiple factors, rank them by contribution.]

Contributing Factors:
1. [Factor and its contribution to the failure]
2. [Factor and its contribution to the failure]
3. [Factor and its contribution to the failure]

What was present before the failure that could have been an early warning:
[Behavioral precursors, adoption metrics, stakeholder signals — what
existed before the failure event that, with a different response, might
have changed the outcome. This is the behavioral autopsy output.]

What was not present that should have been:
[Missing processes, governance structures, monitoring systems, stakeholder
engagements that would have reduced failure risk or enabled earlier detection.]

SECTION 3: WHAT WILL CHANGE — CORRECTIVE ACTIONS
--------------------------------------------------
[For each corrective action: specific description, owner, deadline,
and success criteria. Vague commitments do not appear in this section.]

Action     | Owner       | Deadline    | Success Criteria
-----------|-------------|-------------|------------------
[Action]   | [Name/Role] | [Date]      | [Measurable outcome]
...

SECTION 4: WHAT WAS LEARNED — INSTITUTIONAL KNOWLEDGE
------------------------------------------------------
What this postmortem teaches us about AI initiative design:
[Principles or requirements that should apply to future AI initiatives
based on what this failure revealed.]

What this postmortem teaches us about our organizational readiness:
[Findings about psychological safety, stakeholder dynamics, technical
infrastructure, or governance capability that apply beyond this initiative.]

What this postmortem teaches us about this specific use case:
[Findings specific to this type of AI application — the failure modes
and success conditions that apply to this use case going forward.]

STAKEHOLDER DISTRIBUTION
-------------------------
This postmortem will be shared with: [List]
Executive summary version for: [List]
External communication required: Yes / No
   If yes, audience and key messages: [Details]

FOLLOW-UP SCHEDULE
------------------
30-day corrective action check: [Date] with [Owner]
90-day progress review: [Date] with [Owner]
Postmortem closure (when all corrective actions complete): [Estimated date]