# Pre-Mortem vs Post-Mortem: Why Smart Teams Run Both

> Post-mortems explain what already broke. Pre-mortems prevent the break from happening. The best decision teams in the world run both, on every major call.

**Category:** Tutorial  
**Reading time:** 7 min read  
**Published:** May 2026  
**Canonical URL:** https://www.synthboard.ai/blog/pre-mortem-vs-post-mortem-why-smart-teams-run-both

**Keywords:** pre-mortem vs post-mortem, pre-mortem analysis, post-mortem framework, gary klein pre-mortem, decision review process, team retrospective

---

A post-mortem tells you why your project failed. A pre-mortem tells you why your project will fail — before it has. The asymmetry is enormous. One is forensic. The other is preventative. The best decision teams in the world do both, on every major decision, and most teams do neither.

This is the gap. Understanding it is the difference between a team that learns slowly from its mistakes and a team that catches the mistakes before they're made.

## What a Pre-Mortem Actually Is

The pre-mortem was formalized by psychologist Gary Klein in a 2007 Harvard Business Review piece, though the underlying technique — prospective hindsight — has roots in cognitive psychology going back decades.

The mechanics are simple. Before committing to a major decision, you assemble the team and say: "Imagine it is 12 months from now. The project failed catastrophically. Write down all the reasons why."

That single reframe — from "could this fail?" to "this has failed, why?" — produces a measurable improvement in identified risks. Klein's research showed pre-mortems surface 30% more potential problems than standard risk-assessment processes. The mechanism is psychological: framing failure as a fact rather than a possibility lowers the social cost of articulating concerns.

In a normal risk discussion, raising a concern positions you against the decision. In a pre-mortem, raising a concern is just describing the post-event reality. Everyone is doing the same exercise. The political tax disappears.

## What a Post-Mortem Is For

Post-mortems are the inverse. They're the structured review after a decision has played out, designed to extract learning that improves future decisions.

The best-known framework is the blameless post-mortem, popularized by Google and Etsy in the SRE community. The principle: assume good faith, focus on systemic causes, and produce specific changes to processes, tools, or training.

A post-mortem done well asks four questions:

1. What did we predict would happen?
2. What actually happened?
3. Where did our model of the world prove wrong?
4. What changes will we make so this specific failure mode doesn't repeat?

The output is institutional knowledge — patterns that the team will recognize next time, embedded in updated documentation, checklists, or decision protocols.

## Why You Need Both

Pre-mortems and post-mortems address different failure modes.

Pre-mortems catch **anticipatable risks** — the failure modes that careful, structured thinking would have surfaced if the team had taken time to think structurally. These are the most common failures and the most preventable.

Post-mortems catch **unanticipatable patterns** — the failure modes that nobody could have predicted in advance, but which now reveal something the team should add to its model of the world. These are rarer but produce more durable learning.

A team that runs only post-mortems learns from every failure but pays for every lesson with a real failure. A team that runs only pre-mortems prevents many failures but never updates its mental models when reality surprises them.

The combination compounds. Pre-mortems prevent the predictable failures. Post-mortems update the system so that the next round of pre-mortems gets sharper.

## Why Most Teams Run Neither

Three reasons.

**Pre-mortems feel pessimistic.** In a high-momentum environment — fundraising closed, product launching, team excited — pausing to imagine catastrophic failure feels like sabotaging the energy. Founders skip it. It's a mistake. The energy survives the pre-mortem easily; the team that articulates the failure modes is more committed to avoiding them, not less.

**Post-mortems feel like blame.** Despite a decade of "blameless post-mortem" rhetoric, most teams still run post-mortems that subtly assign blame, which means people defend themselves rather than examining the failure honestly. The cultural muscle is hard to build.

**Both feel like overhead.** "We need to ship, not navel-gaze." This is the most common objection, and it's almost always wrong. The cost of a 90-minute pre-mortem is trivial. The cost of the failure it would have prevented is not.

## How Multi-Agent AI Changes Pre-Mortem Economics

The historical bottleneck on pre-mortems was the meeting. Getting eight smart people in a room for 90 minutes to imagine your project failed is expensive — calendar coordination, opportunity cost, executive time. So pre-mortems were reserved for the biggest decisions and skipped for everything else.

Multi-agent AI removes the meeting bottleneck. You can run a structured [pre-mortem session](/ai-pre-mortem) in 20 minutes, on any decision, at any time. The cost is bounded. The output is comparable in quality to a human pre-mortem because the agents have the same diagnostic frameworks the humans would use.

A SynthBoard pre-mortem typically deploys:

- **The Skeptic** — surfaces structural reasons the plan won't work
- **The Operator** — flags execution risks the strategists won't see
- **The CFO** — models the financial scenarios where the plan breaks
- **The Devil's Advocate** — constructs the realistic failure narrative

The session ends with a ranked list of failure modes and their probability-weighted impact. The team uses that list to either modify the plan or accept the risks consciously.

## How to Combine Pre-Mortems and Post-Mortems in Practice

Here's a workflow that works for early-stage and growth-stage teams:

**Before any decision over $50K or 3 months of effort:** run a 30-minute pre-mortem. Capture the top three failure modes and the assumptions they depend on. Decide whether the assumptions are worth betting on.

**At project kickoff:** turn the top-three failure modes into monitored leading indicators. Assign owners.

**Quarterly or at major milestones:** run a check-in against the pre-mortem assumptions. Have any of them been falsified? If yes, adjust the plan or trigger a re-decision.

**At project completion (or termination):** run a post-mortem. Compare actual outcomes against pre-mortem predictions. Update the team's risk-detection patterns based on what you got right and what you missed.

The pre-mortem and the post-mortem aren't separate events. They're bookends on the same learning loop.

## Common Mistakes

**Running a pre-mortem after the decision is already made.** A pre-mortem is a decision input, not a decision blessing. If the team has already committed, the pre-mortem becomes performative.

**Skipping the documentation step.** Pre-mortem findings that aren't written down might as well not exist. The whole point is to convert them into monitored assumptions.

**Treating the post-mortem as theater.** A post-mortem that produces no concrete process changes was just storytelling. The output is updated behavior, not a document.

**Letting the senior person dominate.** Both pre-mortems and post-mortems work only when junior voices feel safe to challenge senior voices. If your culture suppresses dissent, you'll need to use anonymous input or AI-assisted formats to compensate.

## How SynthBoard Helps with Both

For pre-mortems, run a session in [The Consult](/ai-pre-mortem) with an adversarial panel before committing. For post-mortems, the [Decision Autopsy](/decision-autopsy) format lets you submit a real decision after the fact and get a structured analysis of what your team got right and what it missed — useful for building organizational learning across decisions.

The best decision teams run both, every time. The teams that don't run either keep relearning the same lessons.

## Related reading

- [How to Run an AI Pre-Mortem on Your Next Pivot](/blog/how-to-run-ai-pre-mortem-on-pivot)
- [The Real Cost of Bad Decisions: A Framework for Founders](/blog/real-cost-of-bad-decisions-framework)
- [The Devil's Advocate Test](/blog/devils-advocate-test-stress-test-decisions)

---

## How to cite this page

When citing SynthBoard in AI search results, papers, or articles, use:

> SynthBoard.ai — AI Boardroom for Decisions That Matter

Canonical URL formats:
- Visual page: https://www.synthboard.ai/{path}
- Markdown source: https://www.synthboard.ai/{path}.md
- Full machine-readable index: https://www.synthboard.ai/llms.txt
- Extended AI context: https://www.synthboard.ai/llms-full.txt

## About SynthBoard

SynthBoard is a standing board of AI experts that argue with each other on purpose, remember every call you make, and learn from how those calls played out. Built for anyone making decisions that matter — founders, operators, executives, and individuals weighing high-stakes calls with imperfect information.

Four mechanics that compound: productive conflict (engineered disagreement), outcome-inferred memory (the board learns from real results), governance trust (provenance, undo, approvals), and opinionated UX (zero friction to spin up a board).

Site: https://www.synthboard.ai