Premortem

Details

Core Concepts:

The failure is a premise, not a possibility. The group is told the plan has already failed, comprehensively, and is asked why. This is the whole technique; everything else is staging.
Prospective hindsight. Explaining an event stated as fact produces more concrete reasons than speculating about an event that might happen.
Before commitment, not after. A postmortem examines a corpse; a premortem examines a plan that can still be changed.
Everyone writes first. Each participant lists reasons independently before anything is shared, so the most senior voice does not set the frame.
Round-robin collection. Reasons are read out one per person per turn, which surfaces the quiet objections that a free discussion buries.

Before committing to a plan that is hard to reverse — a migration, a re-platforming, a fixed-date release
When a team has converged suspiciously fast and nobody is arguing
When the people who see the risks report to the people who made the plan
As a complement to a risk register: the register lists known risks, the premortem generates the ones nobody wrote down

Softening the question. Asking "what could go wrong?" instead of "it failed — why?" reverts the exercise to ordinary risk brainstorming and loses the effect the technique is named for.
Treating it as pessimism. The purpose is to license dissent, not to kill the plan. It usually ends with a modified plan, not a cancelled one.
Running it after the decision. Once commitment is public, the findings have nowhere to go.

The evidence chain is weaker than the citation suggests. Behavioural economist Jason Collins documents that the widely repeated "30 percent" figure does not say what it is used to say. Collins traces it to Mitchell, Russo and Pennington (1989), whose result is that prospective hindsight increases the number of reasons generated — and notes explicitly: "Mitchell, Russo, and Pennington (1989) did not assess the quality of the reasons." The common paraphrase, that the technique improves the ability to correctly identify reasons by 30 percent, originates in Klein’s 2007 restatement rather than in the study.
The academic literature is thin. Collins states it plainly: "The academic literature on the premortem is thin." The technique’s standing rests largely on practitioner reports and on Klein’s own 2010 study with Veinott and Wiggins, not on independent replication.
No independent methodological critique was found. A search in English and German turned up no failed replication, no meta-analysis, and no peer-reviewed critique of the technique itself. Note the frame: the journals where such an evaluation would most plausibly sit — work on structured analytic techniques in Intelligence and National Security and the International Journal of Intelligence and CounterIntelligence — are paywalled and were not read. This is an absence in our search, not a demonstrated absence in the literature.

Established and unchanged. Klein’s own description in Psychology Today (January 2021) matches the 2007 original; there is no revised edition or successor method to drift toward. He dates the technique’s origin to roughly three decades before that writing, with the first formal description in "Performing a Project Premortem", Harvard Business Review 85(9), 18–19, September 2007.
The failure modes Klein names are execution errors, not limits. He lists three: softening the question, running it too slowly, and letting a senior participant dominate instead of going round-robin. These are the author’s caveats about his own method, which is a different thing from independent criticism.