A premortem asks you to imagine that a proposed project has already failed, then work backward to plausible causes. The useful output is a revised plan and a few checks, rather than a convincing story about an inevitable failure.
Gary Klein describes the premortem as a team exercise conducted after a plan has been presented: participants imagine failure and identify possible threats. The solo exercise below is our original adaptation for a small personal project. It does not test whether failure will occur or establish how likely a scenario is.
Original fictional case • the proposed plan
Make an audio guide to a community garden
The maker plans to record short descriptions of garden features, upload the recordings, and put a link on a printed sign. The intended result is that a visitor can open the guide at the garden entrance and understand which recording matches each feature.
This definition matters. “Finish the recordings” would count a completed file as success even if visitors could not find or play it.
Imagined future • not an observed event
The guide is finished, but visitors cannot use it
Temporarily take that outcome as the premise. Write several plausible explanations without defending the current plan. Perhaps the sign points to a restricted file; perhaps the link works at home but the audio is hard to hear outside; perhaps the recordings use feature names that do not appear anywhere in the garden.
These are hypotheses to inspect. Do not attach a made-up percentage or treat the most vivid story as the most probable one.
Move from a failure story to a check
Access failure
Imagined cause: the upload requires the maker's account.
Evidence now: the maker has only opened it while signed in. Access for other visitors is untested.
Plan change: before printing the sign, ask a willing tester to open the link while signed out, on their own device. If access fails, repair permissions and repeat the check.
Listening failure
Imagined cause: outdoor conditions make the spoken descriptions difficult to follow.
Evidence now: recordings have only been reviewed indoors.
Plan change: try a sample at the intended location and provide a readable text version. Check that the text can be reached from the same landing page.
Navigation failure
Imagined cause: a visitor cannot match “north border” to a visible feature.
Evidence now: the script uses the maker's private labels; no visitor has tried following them.
Plan change: test a sample description with someone unfamiliar with the project. Revise labels using features a visitor can identify from the entrance.
The access check tests a specific technical assumption. The listening and navigation checks explore use in context. None proves universal accessibility; each can reveal a preventable obstacle before the whole guide is produced.
Choose which changes belong in the plan
Inspect causes that would prevent the defined result and that you can meaningfully investigate before committing more effort. An elaborate redesign is premature if the basic link has never been tested. A speculative concern with no actionable check can remain a note rather than becoming another task.
Assign an owner and a point in the sequence for each retained check, even when the owner is you. “Test the link before printing” can change what happens. “Remember technology problems” is too vague to affect the plan.
The plan after the premortem
- Make one representative recording and a matching text description.
- Test access, listening, and navigation at the intended location.
- Revise the sample before producing the remaining descriptions.
- Print the sign after checking the final visitor-facing link.
These are proposed changes in the fictional case, not completed tests. A real project record should add the date and result when each check occurs.
Premortem, decision journal and after-action review
A premortem examines a proposed plan before the result exists. A decision journal preserves why you chose a course of action at the time. An after-action review compares the intended result with what actually happened afterward. Keep imagined causes in the premortem separate from observations you later record.
Run the exercise on a plan you actually have
Write the intended result and the current sequence. Imagine a clear failure of that result, list possible causes, and separate known weaknesses from unsupported stories. Turn the most consequential, testable causes into checks or revisions. Save the original plan beside the revised one so you can see what the exercise changed.
Stop when you have a manageable set of concrete changes. If the exercise produces only increasingly remote catastrophes, return to the project's boundary and available evidence. The aim is to inspect assumptions early enough to act on them; it is not to remove every uncertainty before starting.
Explore related methods and practical reflection in LUVIRE Guides.