Skip to content

Run a premortem on the proposed research conclusion

Imagine how a recommendation could prove misleading and inspect the specific evidence weaknesses that would produce that failure. Keep hypothetical failures labeled.

Guides

What could make this research conclusion wrong in a useful way?

Before recommending a new help sequence for a home security app, ask how the conclusion could later turn out to be misleading. Perhaps the quoted workflow is obsolete, all examples trace to one incident or the misunderstanding concerns a different task. A premortem converts those plausible failure paths into checks without pretending the imagined failures have actually occurred.

Example research prompt
Run an explicitly hypothetical premortem on supplied security app help recommendations. Check version relevance and source dependence, and retain unresolved assumptions beside the action.
01 / Guides

State the recommendation and assumptions

Write the proposed action and the evidence conditions it depends on. Include scope, version relevance and independent support where needed. Exclude generic pessimism that has no connection to the decision; the exercise should identify plausible ways this particular conclusion could fail, not produce an unlimited list of things that can go wrong in any research project.

02 / Guides

Describe hypothetical failure paths

Label each path explicitly as a scenario, such as all examples describing an older interface. Do not attach invented behavior to real accounts or present a hypothetical correction as an observed event. Keep the scenarios grounded in actual evidence gaps so the exercise remains a focused quality check rather than speculative storytelling that overwhelms the available findings.

03 / Guides

Prioritize concrete checks

Choose checks that could materially change the recommendation, such as reviewing stated versions or tracing source dependence. Use only permitted, available evidence and record blocked checks honestly. If a high-consequence assumption cannot be inspected, make the recommendation conditional rather than inventing reassurance or expanding into unauthorized account access to complete the imagined validation path.

04 / Guides

Deliver a revised conclusion

Report which failure paths were checked, which remain unresolved and how the recommendation changed. Preserve the original observed findings even when the action becomes narrower. A premortem is not a guarantee against error; it is a disciplined way to expose decision-relevant weaknesses before a polished research narrative makes those weaknesses harder for the owner to notice.

Keep the boundaries clear.

  • Hypothetical failure paths guide checks; they are not additional observed research results.
  • Examples are illustrative research prompts; badges are context, not access guarantees or live proof.

Start with a read-only check: ask your agent to report actual source coverage before collecting records or exporting an audience.

Primary references

Documentation reviewed 2026-11-13. Product capabilities, client availability and plans can change. Third-party marks identify their products, not an affiliation.

Know what social media shows about you.
Start with a question.

Get started
montage