Prompt Workshop
Why it is written this way
Tap a block to see its paragraph
You are a post-mortem investigator tasked with analyzing failed projects. Your goal is not to assign blame, but strictly to uncover where things went wrong.
Please assume that the {{time period}} has fully passed. Despite executing according to the plan below, this project was a total failure and fell significantly short of the goal: "{{success criteria}}". I want to read this failure investigation report before we actually start.
Write a detailed investigation report identifying why this project failed. Do not mention any potential for success, silver linings, or positive factors in this response.
For each scenario, structure the breakdown sequentially in the following steps: 1) The exact moment and visible symptom when the failure surfaced 2) What had already unfolded behind the scenes by that point 3) The initial decision or underlying assumption that triggered that chain of events 4) The early warning signs we could have noticed at the time that initial decision was made.
Provide 3 distinct scenarios driven by completely different root causes. Format each scenario as: "One-line Title → The 4 Sequential Steps Above → 2 Preventative Actions We Can Take Right Now". Conclude with a single summary paragraph highlighting the common root cause shared across all three scenarios.
Project Execution Plan: """ {{project execution plan}} """
- Role Who answers ×1
- Material What the AI needs to know ×2
- Work What to do, and how ×2
- Output What comes back, in what shape ×1
The primary goal of a pre-mortem prompt is to eliminate unhelpful optimism. If you hand over a plan and ask, "What do you think?", you usually get three generic compliments followed by a mild "The timeline might be tight." By firmly declaring upfront that the project has already failed, the entire trajectory of the response shifts. An investigator assigned to find root causes has no reason to list positive aspects.
Read the full explanation · 3 paragraphs
The role in the first paragraph is not a consultant or advisor, but a post-mortem investigator. While an advisor tries to offer a balanced perspective, an investigator searches exclusively for fault lines. The context paragraph is the core engine of this prompt. Shifting the tense to "it has already failed" rather than "it might fail" stops the AI from evaluating probabilities and forces it to reconstruct a narrative of events.
The step-by-step framework paragraph forces a trace back from the final symptoms to the very first decisions. Asking open-ended questions about failure causes often results in surface-level excuses like "lack of resources." By enforcing this four-step sequence, the analysis exposes which specific decision triggered that resource shortage—and actionable changes today are almost always found at that initial decision point.
The format paragraph requires three distinct scenarios to prevent the analysis from converging on a single generic issue. When only one scenario is requested, it almost always ends up being a predictable schedule delay. Mandating distinct root causes forces the AI to explore different angles, such as technical debt, human factors, or external dependencies. The concluding summary paragraph reveals overarching patterns across all three narratives. Limiting preventative actions to two per scenario is also intentional; receiving ten action items rarely leads to executing even one.
Unfamiliar terms? See Aha AI: role-prompting, chain-of-thought
Compared with a bad example
What do you think of this project plan? Let me know if you see any issues.
(Paste project plan)
Asking like this typically yields two paragraphs of praise followed by a generic line like "The timeline seems slightly tight." This is far from proactive risk identification. Even if critiques are offered, they usually just echo the phrases written in the plan itself, leaving completely omitted blind spots entirely unaddressed.
Variations
For Running a Interactive Team Meeting
A pre-mortem is originally designed as a collaborative team exercise. Bringing a targeted questionnaire rather than a ready-made report encourages team members to surface the hidden risks visible from their respective roles.
Show prompt
Assume that we executed the plan below over the {{time period}}, but the project completely failed. Generate 10 specific questions for the team to answer together in a meeting.
Instead of open-ended questions like "What went wrong?", phrase each question so it can be answered in a single direct sentence, such as: "Who failed to make what decision by when, causing work to stall?" In parentheses after each question, add the specific role or job function most likely to know the answer.
Project Execution Plan: """ {{project execution plan}} """
For Mid-Project Checkpoints
During mid-project reviews, focus must stay on the remaining time rather than judging past actions. Categorizing warning signs into visible versus hidden makes it clear what needs immediate verification this week.
Show prompt
We started with the plan below and are currently halfway through execution. If we fail to achieve our "{{success criteria}}" within the remaining {{time period}}, outline 3 distinct reasons why that would happen.
For each reason, specify: "Signals that have already appeared · Signals that have not yet appeared · What to verify this week". Do not write retrospective evaluations of past work.
Project Execution Plan: """ {{project execution plan}} """
Model notes
Some models may ignore the assumption of failure and attempt to list success factors alongside risks. If this happens, follow up with: "Remove all success factors and rewrite only the failure investigation narrative."
Last updated 2026-09-02 · Found a mistake? Let us know