This is going to sound strange if you’ve never run a premortem: before a project begins, imagine it’s already finished, and it’s failed. Then work backwards. What went wrong?
I try to run one before every project. Not because I’m pessimistic. Because it’s the easiest way to get people to say the awkward thing in week zero, instead of week two.
Why it works when normal risk conversations don’t
Ask an enthusiastic, ready to get going team “any concerns about this plan?” at kickoff and you’ll mostly get silence. Nobody wants to be the person raising doubts about a project everyone’s just got excited about. It feels negative. It feels like you’re going to be marked down as the difficult one.
Ask the same people to imagine the project’s already failed and explain why, and something shifts. They’re not criticising the plan any more, it’s just a story about what happened in a hypothetical scenario. People say things in that framing they’d never volunteer in a normal “any concerns?” round.
I’ve watched this play out in the room. The same person who says “no, all good” in a normal kickoff will happily write “the client’s approvals process caused a two-week delay” on a post it in a premortem, ten minutes later, in the same meeting. Nothing about the risk changed. Only the framing did.
How I actually run it
Everyone writes down their own answer first, in private, before anyone speaks. If you skip this bit and go straight to open discussion, the first confident voice in the room sets the tone and everyone else quietly agrees with it. Remember it’s a hypothetical situation, different people will have different perspectives as to what could go wrong and you want to hear all the voices.
Then we share. Read them all out, no names attached if that helps. Patterns show up fast – if three people independently wrote some version of “the client’s design sign-off process,” that’s not a coincidence, that’s a real risk with three separate votes.
Then come the actual work and the point of the meeting: which of these are real enough to plan for, and what do we do about them now, not later.
This part matters more than people expect. A premortem with fifteen minutes of writing and discussion, followed by zero changes to the actual plan, has taught everyone that raising a concern doesn’t do anything. Do that once and the next premortem gets quieter, not louder.
What it isn’t
It’s not carte blanche for doom-mongering, and it’s not a box-ticking exercise. It only works if the risks that come out of it actually change something – get written into the Dungeon Master’s Guide, get it raised with the client before a line of code has been written, get it built into the plan. A premortem nobody acts on is just an expensive way of being negative and telling old “war stories” for twenty minutes.
The test I use: for every risk that comes up, someone should be able to say on day one of build, what’s changed in the plan because of it. If nothing’s different, it wasn’t really heard, it was just written down.
Run one before you start your next project. Twenty minutes in week zero is considerably cheaper than the same conversation in week eight, once it’s got a name, a Slack channel and an email trail. Plus, you’ll get to hear some fabulous “war stories” from the old timers in the group.