Asking "what could go wrong" rarely works — it invites generic, hedge-everything answers like "scope creep" or "resourcing issues" that don't point at anything specific enough to act on. Asking "imagine it's six months from now and this project failed — why" gets completely different, more concrete answers, because it skips past the polite hedging straight to a specific story.
The framing change is the entire technique
A premortem isn't a new analysis method — it's the same risk-brainstorming exercise teams already try to do, run with one framing change: certainty of failure instead of possibility of failure. That shift matters because hedged language ("this might be a risk") lets people stay vague and non-committal, while a concrete failure story forces specifics: which dependency broke, which assumption turned out wrong, which stakeholder pulled support.
It works because it removes the social cost of pessimism
In a normal risk discussion, being the person who keeps raising doubts about a project everyone's excited about carries a real social cost — it reads as not being a team player. A premortem removes that cost by making pessimism the explicit task for everyone in the room, for ten minutes, instead of one person's unwelcome opinion. People who'd never volunteer a doubt unprompted often produce the sharpest failure stories once asked directly.
Run it early enough that findings can change the plan
A premortem run after the plan is locked and resourced just produces a list of risks nobody has the appetite to act on. Run before commitment, while the plan can still shift, the same exercise actually changes what gets built and in what order — which is the entire point of finding the failure mode in advance instead of in the postmortem six months later.
Don't stop at the list — pick the two stories worth planning against
A premortem that generates twenty failure stories and addresses none of them is no better than skipping it. The output that matters is picking the two or three most plausible, most damaging stories and asking what change to the plan would have prevented them — turning a brainstorm into a small set of concrete adjustments, not just a longer risk register nobody revisits.