Every leadership team I have sat in has had a moment, a few months into a big decision, when somebody says "I had a feeling about this". They did have the feeling. They did not say it, because the meeting where the decision was made had already tilted toward yes, the CEO was visibly pleased, and raising a doubt would have meant being the person who slowed things down. The premortem exists for that person and that moment. It moves the doubt from three months after the decision to forty minutes before it.
What a premortem is
Gary Klein described the exercise in the Harvard Business Review in 2007, and the shape has not changed since. The team gathers once the plan is drafted and before it is final. The leader says: it is a year from now, we did this, and it failed. Badly. Then everyone writes down, on their own and in silence, every reason they can think of for the failure. After a few minutes, the room goes around, one reason per person per turn, until the lists run dry. The leader listens and writes.
That is the whole thing. No facilitator, no slides. Klein's version takes under an hour, and most of the value is in the five silent minutes.
Why the research says it works
The mechanism is called prospective hindsight, and it was measured before Klein gave it a name. Mitchell, Russo and Pennington (1989) asked people to explain a future event either as something that might happen or as something that had already happened. Treating the outcome as certain raised the number of reasons people produced by roughly 30 percent, and the reasons were more specific. Asking "what could go wrong?" invites a shrug. Asking "it went wrong, why?" invites a story, and stories carry details.
Veinott, Klein and Wiggins (2010) tested the premortem directly against a plain list of pros and cons and found it cut overconfidence in the plan by about twice as much. Daniel Kahneman, who spends a chapter of Thinking, Fast and Slow (2011) on the limits of every debiasing technique he knows, singles out the premortem as the one he would keep. His reason is that it legitimizes doubt. Once the leader has said "this failed", saying why is loyalty rather than dissent.
The problem with dissent in a small company is rarely that people have nothing to say. It is that saying it costs them something, and the cost goes up the more enthusiastic the boss looks. The premortem lowers the cost to zero for one meeting.
The two mistakes that make it a formality
I have seen the exercise fail twice in the same way, and both times it was the same mistake wearing different clothes.
The first is skipping the silent writing. Somebody suggests going straight around the table, the first confident person speaks, and the next four reasons are variations on theirs. The whole point of writing alone is that the quiet analyst's reason gets onto paper before the loud one's reason fills the room. Stasser and Titus showed in 1985 that groups mostly discuss what everyone already knows and leave the unshared facts on the table; the silent minutes are the cheapest fix for that anyone has found.
The second is running it too early or too late. Too early, before there is a plan, and the causes are generic ("the market shifts"). Too late, after the announcement, and nobody will write a reason that implies the CEO was wrong. The window is narrow: the path is chosen, the plan is drafted, the commitment has not been made.
What to do with the list
A premortem that ends with a list on a whiteboard has done half its job. The other half is turning each cause that matters into something you can check. "The new hire does not work out" becomes "we will know by the end of month two whether she has closed a deal on her own". "Customers balk at the price" becomes "if churn in the first quarter exceeds what we lose today, the price was wrong".
Written that way, a cause is a condition: a statement that would have to be true for the plan to work, with a way to find out. Some of those conditions you can check now, by asking the person who would know. Some become tripwires, the signals that mean the decision needs another look. Either way, the list has stopped being a feeling and started being a record.
Where Midfire fits
Midfire runs the premortem as a step on every big bet, with the silent minutes built in: each person asked writes their causes without seeing anyone else's, and the causes land as conditions on the decision's record, where the team can answer them and the look-back can score them. The exercise is Klein's; the product's contribution is making the honest version the default one.
The Monday version
Take the largest decision on your desk right now. Write the path you are leaning toward in three sentences. Send those sentences to the four people whose read you would want, with one instruction: it is next September, this failed, write me five reasons, and do not talk to each other first. You will have your premortem by lunchtime, and at least one of the reasons will be something you had not thought of. In my experience it is usually the one from the person who speaks least in meetings.