The Offer Expires at Midnight
Mara had ninety minutes.
The offer letter for Devraj sat open on her laptop, the cursor blinking in the salary field. Six weeks of interviews. The board loved him. She loved him — the way he'd torn apart their pricing model in the final round like he'd already worked there. The VP of Engineering role was his if she hit send before the candidate's other offer expired at midnight.
She told herself the only open question was the number.
That was the lie, and she almost sent it anyway. It's what you do at 10:30 p.m. on a Tuesday when you've already decided and the only thing left is paperwork.
Then she remembered the line from the audit she'd run that morning, half as a joke: If this already failed, what did I miss?
She didn't want to ask it. Asking it meant admitting the thing she'd been holding back all along — that Devraj's brilliance had been on their problems. They'd handed him the model. They'd framed the questions. The real job, the one the board expected him to own in month two, was to build the strategy when nobody had handed him the script. They had never once watched him do that.
She wrote it down anyway. "We hired the person who solved our case. We never tested the person who'd have to find his own."
The pre-mortem made her honest, because once failure was assumed, naming the risk stopped being disloyalty and started being the competent thing to say. She sorted it against the five ways plans die: not adversarial, not confidence — a World Model Gap. She'd modeled Devraj against the role they wrote, not the role the quarter would force.
The cheapest move wasn't to withdraw. It was a thirty-day cross-team trial before the offer converted — two weeks of him working the actual gap, in front of the people who'd report to him. Cost: a paragraph in the email. Saved: the €120K, the team, and the year.
She changed the send. Not “accepted.” "Offer contingent on a 30-day working trial — here's the shape of it."
Devraj replied in four minutes. "Fair. I'd have asked for the same if I were you."
At midnight, the other offer expired. Mara's was still open — but now it was a decision she'd actually made, not a default she'd rubber-stamped. The side she couldn't see from where she was standing had been shown to her, before she committed.
That's what DAUDIT is for. Not to tell you no. To give you the question your own mind was saving for last.
---
Most plan reviews ask the room to defend the plan. Everyone reaches for the argument that keeps them in the room. The objection that would have saved you never gets spoken, because speaking it reads as not being a team player.
The pre-mortem flips that incentive. You do not defend the plan. You assume the decision is already made, the date is six months out, and it went wrong — completely wrong. Now write down why.
That is the whole technique. You are not predicting the future. You are giving your mind permission to say the thing it has been holding back.
Why it works
A normal review rewards defense. A pre-mortem assumes failure is settled, so the only way to look competent is to name the real risk first. The person in the corner who stayed quiet suddenly talks. The risk you dismissed as “unlikely” gets a face.
Gary Klein described this as prospective hindsight in 2007. In his study, groups running a pre-mortem found on average roughly twice the number of genuine failure paths as groups doing a standard review. Not because they were sharper. Because the format let them.
The five ways plans actually die
A good pre-mortem forces each failure into the light. They cluster into five dimensions — the same ones DAUDIT examines on a live decision:
- World Model Gaps — what your picture of the situation leaves out. The market moved. The customer you modeled isn't the one who shows up.
- Adversarial Risk — who benefits from you being persuaded. The counterparty is not optimizing for your outcome.
- Theory of Mind — what the other parties actually believe, not what you need them to believe.
- Decision State — whether this decision is still yours to make, or whether momentum, sunk cost, and a calendar already made it.
- Confidence vs Evidence — where your certainty sits relative to what is actually true.
The pre-mortem names the failure. The five dimensions tell you which kind it is, and what to do before you are standing inside it.
How to run one without a facilitator
You do not need a workshop. You need fifteen minutes and the honesty you normally withhold from yourself.
- Set the date. “It is six months from now. We committed. It failed.”
- Write the obituary. Not bullet points — a short, specific story of what went wrong. Specificity is the test. “Culture fit” is not a failure. “The senior hire resigned at month three and took the roadmap” is.
- Sort each cause into the five dimensions above. One will dominate. That is your blind spot.
- Decide the cheapest move that removes or insures against it — before you sign.
The part people miss
A pre-mortem is not a one-time gate. The decisions that actually sink you are the irreversible ones, and they rarely arrive with a flag. They arrive looking like the obvious next step.
So the exercise is less about any single call and more about building the reflex: before I commit, let the failure talk.
DAUDIT's Pre-Mortem exists to run this on your actual decision — across the five dimensions, naming the angle you didn't think to look for. The Opening Read is free and anonymous. The full audit needs an account so the record persists and you can see, a year later, which blind spots you caught in time.
But you do not need the tool to start. You need the question:
If this already failed, what did I miss?
Write the answer down before you commit. That is the whole technique.