What's the Point of a Post-Mortem?

Why the best reviews look beyond mistakes to uncover what really happened and what needs to change.

The point of a post-mortem is to understand what happened, why it happened, and what should change as a result. Whether a project failed, a product succeeded, or a team missed an important deadline, a post-mortem examines the decisions and conditions behind an outcome so people can make better decisions next time.

A good post-mortem separates what people knew at the time from what became obvious afterward. It identifies contributing factors rather than settling for the easiest explanation.

That distinction matters because organizations can document an event accurately and still learn the wrong lesson.

The Danger of Stopping at the Obvious Explanation

On February 28, 2017, Amazon Web Services experienced a major disruption to its S3 cloud storage service in Northern Virginia.

An authorized employee, following an established procedure while investigating a billing system problem, entered an incorrect input into a command. The command removed far more servers than intended, disrupting systems essential to S3 operations.

The immediate explanation was straightforward: someone entered the wrong information.

But that explanation alone would have missed the larger problem.

According to AWS's published incident summary, the operational tool allowed too much server capacity to be removed too quickly. Recovering the affected systems also took longer than expected.

AWS reported modifying the tool to remove capacity more slowly and adding safeguards to prevent commands from reducing critical systems below minimum capacity requirements. It also identified improvements needed in recovery procedures and incident communications.

The important lesson was not simply that an employee should be more careful. It was that a critical system should be designed to limit the consequences of predictable human mistakes.

That is what an effective post-mortem can reveal.

A status report describes the disruption. A post-mortem examines the conditions that allowed it to happen.

The difference is consequential. Training an employee to enter commands more carefully addresses one person's behavior. Changing the system so an incorrect command cannot remove dangerous amounts of capacity addresses a vulnerability that could affect everyone.

Why Blame Can Prevent Understanding

One of the most difficult aspects of a post-mortem is separating accountability from blame.

Blame focuses attention on who made the mistake. Accountability asks what responsibilities people had, whether they fulfilled them, and what needs to change.

Those questions overlap, but they are not interchangeable.

Google's Site Reliability Engineering guidance describes a blameless approach to reviewing incidents. Rather than treating individual error as the end of an investigation, teams examine the technical and organizational conditions that contributed to the outcome.

The reasoning is practical.

When employees expect mistakes to trigger punishment or embarrassment, they have reasons to defend their decisions, minimize uncertainty, or withhold uncomfortable information.

Yet a post-mortem depends on precisely the information that defensiveness can conceal.

Consider what happens when a manager asks, "Who approved this decision?" before asking what information was available when the decision was made.

The first question can establish responsibility. But if it dominates the discussion, the team may never examine whether the approval process itself was flawed.

Blameless does not mean consequence-free. Deliberate misconduct, reckless decisions, and repeated disregard for established safeguards still warrant scrutiny.

The distinction is that accountability should follow an examination of the circumstances, not replace it.

Leaders face a particular challenge here. They must create conditions where people can acknowledge mistakes while maintaining meaningful standards of responsibility. That includes examining management decisions and organizational incentives, not just the actions of employees closest to an incident.

A review that protects leadership from scrutiny while examining everyone else's mistakes is unlikely to uncover the whole story.

Success Deserves Investigation Too

Post-mortems are usually associated with failure. But successful outcomes can conceal weaknesses just as failures can reveal them.

Consider a hypothetical company that completes a major project ahead of schedule.

Leadership celebrates excellent planning and execution. A closer examination reveals that several employees worked unsustainable hours, a supplier delivered early through luck, and one experienced employee quietly prevented problems that normal procedures would not have caught.

The project succeeded, but the process was fragile.

This illustrates an important distinction: the quality of a decision is not the same as the quality of its outcome.

A sound decision can produce disappointing results because of circumstances beyond anyone's control. A poor decision can produce excellent results through favorable circumstances.

Judging decisions entirely by their outcomes makes it difficult to distinguish good judgment from good luck.

Examining success also reveals practices worth preserving. An informal collaboration between departments, for example, may have prevented delays. Understanding how that collaboration worked could help other teams.

The objective is not to diminish an achievement. It is to discover which parts of the achievement are worth repeating and which depended on circumstances that cannot be counted on.

What Makes a Post-Mortem Useful?

Many post-mortems fail not because their conclusions are wrong, but because they are written to be defensible rather than useful.

A document can contain an accurate timeline, carefully worded observations, and agreeable recommendations without challenging a single important assumption.

Statements such as "communication should improve" or "teams need greater alignment" sound reasonable. They rarely explain what actually needs to change.

A useful review begins with five questions:

1. What did we expect to happen?
2. What actually happened?
3. What explains the difference?
4. What should we preserve, change, or stop doing?
5. Who owns the next steps, and what evidence will show they worked?

The third question demands particular discipline.

Once an outcome is known, warning signs appear more obvious than they were at the time. A useful post-mortem reconstructs what participants actually knew, which alternatives they considered, and what uncertainties they faced.

This helps distinguish decisions that were unreasonable when made from reasonable decisions that produced unfavorable results.

The fifth question prevents recommendations from disappearing into a document nobody revisits.

For a change that can be implemented quickly, schedule a review 30 days later. Ask what changed, whether the change was implemented, and what evidence suggests it is working. More complex changes need longer follow-up periods.

A recommendation without an owner or a way to evaluate progress is a conclusion, not an improvement plan.

The Real Test of a Post-Mortem

Experience does not automatically produce learning.

A team can experience the same failure repeatedly while producing increasingly detailed explanations of why it happened.

A post-mortem interrupts that pattern when it exposes assumptions that previously went unquestioned.

It also requires intellectual humility. Evidence can be incomplete, recollections can conflict, and several contributing causes can remain plausible. An honest review acknowledges those limits rather than manufacturing certainty.

The most useful post-mortem is not necessarily the one that uncovers the most problems. It is the one that changes how people understand the decisions they made and helps them make better ones next time.

🧬 QuestionStrings to Practice: From Experience to Understanding

Use this sequence after an important project or unexpected outcome.

What did we expect to happen?
→ Where did reality differ from our expectations?
→ What evidence best explains that difference?
→ Which assumption or decision deserves reconsideration?
→ What specific change will we test next?

The question worth returning to after every post-mortem is simple: What do we now understand that we did not understand before, and what will we do differently because of it?

Explore more questions at QuestionClass's Question-a-Day.

← Home