Running a Premortem Before a Big Commitment

Why this matters

By the time you are ready to sign for a second truck, a new building or a big hire, you have mostly decided, and everyone around you can tell. The office manager has doubts but you seem sure. The apprentice sees a problem but it is not his place. The questions that would have saved the plan stay unasked, then show up a year later as the reasons it failed. A premortem is a 45-minute meeting that makes those reasons speakable before you commit: everyone imagines the plan has already failed and writes down why. It does not decide for you. It changes the plan you sign.

The idea in one paragraph

A postmortem asks why something died after it is dead. A premortem, the method Gary Klein described in the Harvard Business Review in 2007, asks the same question while the patient is still alive. The trick is the tense. "What could go wrong?" invites polite, hedged answers. "It is a year from now and this failed. What happened?" gives people permission to tell a story, and stories surface specific causes. Klein drew on research by Mitchell, Russo and Pennington (1989) finding that imagining an event has already happened improved people's ability to identify reasons for it by about 30%.

Use it for one-way doors: commitments that are expensive or slow to reverse. A price test does not need a premortem. See related: Reversible and Irreversible Decisions in a Small Shop.

The worked case: a second truck

The steps below are run on one real-shaped plan. A plumbing shop has one truck run by the owner with an apprentice, and the owner's spouse runs the office part time. In April the backlog hit 3 weeks. The plan is to hire a licensed tech and put a second truck on the road. Four people attend: the owner, the office manager, the apprentice and the shop's outside bookkeeper.

Step 1: Write the commitment down with its numbers

Before the meeting, the owner writes one paragraph describing exactly what is being committed to and what success looks like, with numbers.

Truck 2 on the road by June 1, financed over 60 months. One licensed tech hired. Target 28 billable hours a week on truck 2 by September. The bookkeeper puts truck 2's break-even (truck payment, wages, insurance, fuel) at about 22 billable hours a week.

Why: people cannot imagine the failure of a vague plan. "Grow the business" has no way to fail. What breaks if you skip it: each person premortems a different plan in their head, and the reasons do not connect to anything you can change.

Step 2: Set the scene in the past tense

Read it out: "It is June next year. Truck 2 has been sold back and the tech is gone. The plan failed. Take five minutes and write down every reason you can think of for why it failed."

Why: the past tense is the whole method. It turns doubting the boss into an assigned task. What breaks: if you ask "any concerns?" you get none, because concerns sound like disloyalty and the owner has visibly decided.

Step 3: Everyone writes alone, in silence, for five minutes

Paper or phones, no talking, at least three reasons each.

Why: if people speak first, everyone anchors on whatever the owner or the most confident person says, and the apprentice's observation never surfaces. Silent writing gets the independent views on paper before the group can merge them. What breaks: a meeting where the owner opens with "I think the main risk is hiring," and all four lists come back about hiring.

Step 4: Read them out round robin, owner last, no debate

Go around the table one reason per person per round until every list is empty. The owner reads last. Nobody argues, defends or solves yet; someone writes each reason on a whiteboard. A round from this shop went like this:

OFFICE: The new guy's trucks were full in May and
  empty by October. We paid him to sit.
APPRENTICE: I quit in August. He got the good
  calls and I got drains.
BOOKKEEPER: Winter cash. Payment plus wages with
  half the hours. Line of credit maxed by Feb.
OWNER: Couldn't find a licensed tech. Truck sat
  in the lot all summer.
OFFICE: Double bookings. I can't dispatch two
  trucks off a paper calendar.

Why the owner goes last: the owner's reasons carry weight, and once spoken they frame everything after. Why no debate: the moment a reason is challenged, the next person self-censors. What breaks: a 45-minute meeting turns into the owner explaining why each concern is handled, and people stop offering them by round two.

Step 5: Consolidate, then rank by likelihood and damage

This shop produced 17 raw reasons. Several were the same failure in different words; the bookkeeper's winter cash squeeze, for instance, is the same mechanism as the office manager's empty-by-October truck, so they were merged. That left 7 distinct failure stories. Each got a likelihood score and a damage score from 1 (low) to 3 (high), agreed by the group, and the two were multiplied.

The seasonal story scored highest because the bookkeeper brought a number nobody had looked at: in each of the last two Octobers, the backlog was 2 days, not 3 weeks. The April backlog that started this whole plan was a seasonal peak.

Why rank: a list of 17 worries paralyzes; a ranked list of 7 tells you where the plan needs to change. What breaks: the group spends its remaining time on the vivid but low-scoring story (the apprentice walking out) and never touches the one that will actually sink the truck.

Step 6: For the top three, change the plan, name an owner, set a trigger

Each of the top stories gets three things: a change to the plan, a person responsible, and an early-warning signal with a date. Here, the top three:

• Seasonal starvation. Change: before truck 2 launches, sell a fall drain-and-water-heater maintenance plan to existing customers, aiming for 120 signups by September 1, to put booked work into October through February. Lease the truck over 24 months instead of financing over 60. Owner: office manager for the plan sales. Trigger: truck 2 under 18 billable hours a week for 3 straight weeks after September 1 means the owner calls the meeting to park or redeploy it.
• No hire. Change: do not sign for the truck until a tech has accepted. Owner: the owner. Trigger: no accepted offer by May 15 means the launch slips to next spring and the summer is covered with overtime and a sub.
• Owner's hours collapse. Change: the office manager takes over the scheduling and parts ordering the owner was doing at night, and the owner's billable hours are tracked weekly. Trigger: owner under 28 billable hours a week for 4 straight weeks means a review of what he has picked up.

Each trigger is a number and a date because a feeling ("it seems slow") will be explained away by the time it arrives. The triggers are kill criteria in miniature. See related: Setting Kill Criteria Before You Start Something New.

Step 7: Decide whether the plan still stands, then write it down

Ask the plain question: knowing these stories, do we still do this? Here the answer was yes, with a different plan than the one that walked into the room: hire first, lease over 24 months, launch with a fall maintenance plan behind it. The owner recorded the plan, the three triggers and his confidence in a journal entry with a review date of December 1. See related: Keeping a Decision Journal.

Sometimes the honest answer is no. If the top story cannot be mitigated by anything you are willing to do, the premortem has done its most valuable job.

How the meeting itself goes wrong

These are failures of the meeting, not the plan:

• It becomes a grievance session. The apprentice's story turned briefly into a complaint about who gets which calls. Note it, park it for a separate conversation, keep moving. The premortem is about this plan.
• Scores are set by the owner alone. If the owner assigns all the likelihoods, the ranking reflects his blind spots. Agree them as a group, or average independent scores.
• The list is filed and forgotten. If the triggers are not on a calendar with a name next to them, the meeting was theater. The December 1 review is when you check them.
• Everything gets a mitigation. Mitigating all seven stories adds so much cost the plan becomes worse than the original. Three is usually right; the rest get noted and watched.
• It is held after the deposit is paid. Once money is committed, people unconsciously defend the plan. Run it before signatures.

When not to bother

Skip the full meeting when the decision can be undone in weeks at low cost, or when you are the only person involved. For a solo decision, the solo version still works: write the past-tense paragraph, then list your own five reasons before you sign.

References

• Gary Klein, "Performing a Project Premortem," Harvard Business Review, September 2007
• Deborah J. Mitchell, J. Edward Russo and Nancy Pennington, "Back to the Future: Temporal Perspective in the Explanation of Events," Journal of Behavioral Decision Making (1989)
• See related: Reversible and Irreversible Decisions in a Small Shop, Setting Kill Criteria Before You Start Something New
• See related: Keeping a Decision Journal, Groupthink in Partnerships and Management Meetings