Premortem · Pre-Mortem Analysis
Premortem
What is a premortem?
- Assuming failure makes risks easier to surface
- Writing silently and individually prevents groupthink
- Unlike a post-mortem, it happens while you can still change course.
What a premortem really changes is not the length of the risk list but who is entitled to speak. In an ordinary meeting, 'I don't think that will work' is one person's opinion and gets rebutted at once. In a session where failure is already the premise, the same words are received as analysis. So information that practitioners knew but never raised, such as an operating team's peak-season schedule, signs that a key employee may leave, or a partner's delivery track record, makes it into the minutes for the first time. The goal is not to dampen optimism but to turn it into something verifiable, and the person who names a risk becomes its owner as a bonus.
It usually takes one of two forms. The 20–30 minute short version, tacked onto a kickoff meeting, gathers 5–8 people and aims to capture just the top three risks; for large investments or initiatives that are hard to reverse, a half-day extended version is used. Either way, a room of only the people who made the plan yields little. Mix in the field teams who will carry out the work, departments with conflicting interests, and people who have seen a similar project fail, and new ideas surface. It also differs from a risk register. A register is paperwork that fills in fixed items, while a premortem produces failure scenarios that are not yet in any document. Where the threats in a SWOT analysis look at the external environment, a premortem looks at the mistakes we ourselves are likely to make.
The most common point of failure is a single remark from the decision-maker. The moment an executive cuts in mid-discussion with “We already looked into that,” the meeting turns into a defense of the plan, and no one speaks up on the next project. The facilitator should get everyone to agree in advance to write things down first, without judging or explaining on the spot. Another misconception is that the job is done once you have a list. If none of the top items changed the schedule, budget, or scope, that premortem was merely a formality for sign-off. Bolting it onto a plan that is already approved and can no longer be changed, or running it out of habit at every meeting, ends the same way.
John KimAuthor of B2B Marketing Design and CEO of Nanum Consulting
Okgum LeeExpert in corporate analysis, fundraising, and business planningProposed by psychologist Gary Klein and popularized by Harvard Business Review (2007). It builds on research showing that people come up with more concrete causes when they imagine a future event as having already happened.
Consider a hypothetical case. Company A, a Korean auto parts maker, has 620 employees and annual revenue of KRW 240 billion (≈ US$179,560,100). It was about to start an eight-month, KRW 900 million (≈ US$673,400) project to roll out a manufacturing execution system (MES) across all plants, but a similar attempt three years earlier had been halted after six months by pushback from the shop floor. Two weeks before launch, the PMO scheduled a premortem ahead of the kickoff.
- Team lead Kim, who runs the PMO, opened the workshop by stating flatly: 'A year from now, nobody uses this MES and the shop floor has gone back to Excel.'
- Fourteen people from production, quality, IT, and purchasing each wrote five lines in 8 minutes, and the night-shift foreman wrote 'Where are we supposed to find 20 minutes to enter data on a tablet during shift handover?'
- Going around the room, the group read out 41 failure reasons, merged duplicates into 17, and ranked the top 3: non-use on the shop floor, poor master data, and mismatched partner codes.
- The IT team and the production control team argued for 30 minutes over “data cleansing isn’t our job,” and in the end they assigned two dedicated cleansing staff for three months and added KRW 32 million (≈ US$23,900) to the budget.
- Eight months after launch, on-site data entry compliance was 94%, the stabilization period was six weeks shorter than planned, and the PMO reported that an extra KRW 32 million (≈ US$23,900) had prevented KRW 180 million (≈ US$134,700) in rework costs.