Pre-mortem

Where it came from

Gary Klein is a psychologist who spent his career studying how firefighters, nurses and soldiers make decisions under pressure, and he noticed that once a plan existed, the people on it stopped looking for its flaws. He published the pre-mortem in Harvard Business Review in September 2007 as a short, almost trivially simple fix. Tell the team the plan has failed, and ask them why.

It rests on a 1989 experiment by Deborah Mitchell, Jay Russo and Nancy Pennington. They asked people to explain a future event. Some were told it might happen; others were told it had happened, and to explain it in hindsight. The second group came up with about thirty per cent more reasons, and more specific ones. They called it prospective hindsight. Certainty, even pretend certainty, unlocks reasons that possibility doesn’t.

Daniel Kahneman gave the pre-mortem a page of Thinking, Fast and Slow and has recommended it as his preferred check on overconfidence. It costs almost nothing, which is why it is worth running on almost everything.

Why it works

Doubt feels disloyal. Once a group has agreed a plan, pointing out how it might fail sounds like an accusation against whoever proposed it, so people don’t. The pre-mortem makes the failure a premise instead of a prediction. You aren’t saying it might fail. You have been told it did, and your job is to be the person who explains why. Everyone gets permission to be the pessimist at once.

The past tense matters. "We ran out of money" comes out of a different part of the head than "we might run out of money". So does the silent writing: if the room talks, the first confident voice sets the list and everyone else adds variations. If everyone writes alone first, you get the quiet person’s reason, which is often the one nobody else saw.

When to reach for it, and when not to

Reach for it when a plan exists and hasn’t started. That is the window. Run it too early and there is nothing to break; run it once the work is under way and it turns into a list of excuses for what is already going wrong.

It is also useful on a decision you have nearly made. Tell yourself you chose the thing and it went badly, and see what comes to mind. If nothing does, that is worth knowing too. Greggle suggests it first on a Goal, a Project and a Risk, and on a Decision and an Opportunity as well: five of the seven kinds.

How to run it

With a pen, on one sheet of paper.

State what you are about to commit to, in a few lines, and when you would know whether it had worked.
Announce it. "It is a year from today. This went badly. Not slightly, badly." Say it as a fact.
Everyone writes, alone and in silence, every cause they can think of, in the past tense. Two minutes. Nobody talks.
Go round the room. Each person reads one cause. Keep going round until the lists are empty. No discussion yet, and no defending the plan.
Judge each cause against the others: how likely is it really, and how bad would it be? Fatal means the whole thing fails. Serious means a major delay or cost. Annoying means you’d recover.
For each cause worth acting on, write something you could start this month, and say whether you are preventing it, reducing it, or accepting it. An accepted risk gets its reason written beside it.

There is a one-page card with these steps and a blank grid to fill in, made to print on A4. Or the PDF.

On a Greggle board

Pick Pre-mortem on a Project board and Greggle asks what you are committing to and when you would know, then proposes two steps: why it failed, with four causes to start, and what to do about it now, with three actions to start. Both take up to twelve.

A stall at the Saturday market, inside why it failed: four causes in the past tense, one rated very likely and serious — a real board · the market stall, inside why it failed · click a step to open it

Inside 1 — Why it failed: the board one level down — a real board · the market stall, inside why it failed · click a step to open it

the planWhat you are about to commit toAnd when you would know whether it had worked.

1Why it failedIt is twelve months from now. This went badly. Looking back, what caused it? Four causes to start.

each causeWhat went wrongPast tense. How likely is it really, against the others? How bad: fatal, serious or annoying?

2What to do about it nowEach cause worth acting on becomes something you can start this month. Three to start.

each actionWhat to do, and which cause it addressesPrevent it, reduce it, or accept it. If accepting, why.

not on the boardThe silent two minutesThe app lays out the pieces. The rule that nobody talks first is yours to keep.

Nothing is added until you have looked at it. Open any step and it becomes a board of its own. Steps that cannot sensibly be cut smaller get marked as actions, and every action collects in one flat list.

What the check looks for

One button reads the whole branch back. On this method it reports, among other things:

The failure has been imagined and nothing has been written under what to do about it.
None of the causes would actually sink this. The awkward cause is the one that got left out.
All the causes carry the same likelihood, which is a form filled in rather than a judgement made.
A cause is accepted and does not say why. A month from now it will look identical to one nobody noticed.

A worked example

a Project · a stall at the Saturday market

The plan. Rent a stall for four Saturdays in November selling the baked things that sell well at school. Split the profit three ways. We would know by the end of November: did it cover the stall fees, and are we still friends.

The news. It is next spring. The stall lost money and two of the three of you aren’t speaking.

CausesIt rained on two of the four Saturdays and nobody came. We baked too much on the first day and threw half away. Only one of us could actually get there at 6am with the trays. We never agreed what to do if someone couldn’t make it. The council wanted a food hygiene certificate we didn’t have. The person whose kitchen we used got fed up with the mess. The pricing was guessed and we sold out of the cheap thing and nothing else.

JudgedThe 6am problem: very likely, serious. The hygiene certificate: possible, fatal, because no certificate means no stall. Baking blind on day one: very likely, annoying. The unspoken sharing rule: possible, fatal to the friendship. The rain: possible, serious.

ActionsCheck the council’s rules before paying for the stall (prevents the certificate). Do the first Saturday with half the stock and count what sells (reduces the waste). Write down, now, who does what if one of us is ill, and what happens to their share (prevents the row). Ask the one who can do 6am whether they actually want to, and pay them for it (reduces the 6am problem).

AcceptedThe rain. Reason: we can’t move the market indoors, and two dry Saturdays out of four should still cover the fees. If the first Saturday’s takings don’t cover the stall fee, stop after the second and talk before the third.

Notice that the reason most likely to end friendships was the one about sharing, and it took a pretend failure to say it out loud. Notice too that accepting the rain, with the reason written down, is a real answer. Never mentioning it is not.

Where it falls down

Done as a discussion instead of silent writing first, the loudest voice sets the list. Done by one person alone, it still works, but you only get one person’s blind spots. Done too late, it becomes a post-mortem with extra steps.

The other failure is stopping after the list. A pre-mortem that produces twelve causes and no changes to the plan was a nice conversation. The value is entirely in the actions at the end, each one naming the cause it answers, and in the honest decision to accept the causes you can’t prevent, with the reason written down.

Pairs well with

Working Backwards. Write the ending first, plan back to now, then run a pre-mortem on the plan. One walk forward and one back.
Scenario Planning. When the causes of failure are outside your control, the weather or the council, build the futures instead of trying to prevent them.
5 Whys. A cause that surprises you deserves five whys of its own.