Before you spend, tell the AI the project already failed

sweetmelon team · Management

There is money about to go out the door. You ask what it thinks of the plan and get a balanced page of upsides and risks, and you spend the money exactly as you were going to. Change the premise. Paste in the plan, the budget, the timeline and what you expect to get back, then tell it the project is over and it failed completely. Have it work backwards to the causes and rank them by how fatal each one was. A postmortem has to name specific deaths, and those you can judge for yourself. The ranking is the half people skip. Left to itself it will set "the budget may run over" beside "nobody actually wants this" as comparable exposures. Then ask which item on that list is the assumption the whole plan rests on, the one where being wrong makes the rest irrelevant. Asked after the failure analysis, that question gets a better answer than asked cold, because now there is a list to choose from. Last, ask what would change its assessment. That answer is the most useful of the three, because it turns an opinion into errands. Two days of asking around and you know whether the worst item is real. Ignore any probability of success it offers, and do not ask for one. The number arrives with two decimal places and no basis whatsoever. What you came for is the list of what to go and check before the money moves.