The CEO of a sixty-person firm told me last year that his team had spent three meetings on whether to switch expense software and forty minutes on whether to open an office in a second city. I asked why. He said the software decision had a spreadsheet and the office decision had a feeling. That is the whole problem with how most small companies allocate their deciding: the effort goes where the data is, not where the stakes are.
The fix is a sorting step that takes thirty seconds and puts the process where it belongs.
The two questions
Jeff Bezos put the sort in his 2015 letter to shareholders. Some decisions are one-way doors: consequential, hard or impossible to reverse, and they deserve to be made slowly, with deliberation and consultation. Most are two-way doors: if you walk through and do not like what you see, you walk back. Those should be made fast, by individuals or small groups, and the failure mode of a growing company is that it starts using the heavy process on everything.
I would add a second axis, because reversibility is not the whole story. Some reversible decisions are very expensive to reverse. So the two questions I ask on every decision that reaches me are: how much does it cost if we are wrong, and how hard is it to undo. High on both, and it is a big bet. Low on either, and it is an everyday call.
McKinsey's decision research (De Smet, Jost and Weiss, 2019) uses a similar cut with more categories: big bets, cross-cutting decisions that touch several teams, delegated decisions, and the ad hoc ones. For a company under a hundred people I find two categories are enough. The taxonomy matters less than the habit of answering the two questions before you decide how to decide.
Why the big bets need the treatment
Paul Nutt followed more than four hundred real decisions in organizations for two decades and published the pattern in 1999. About half of them failed, meaning they were not sustained in use two years later. The single most reliable predictor of success was whether the decision-maker had seriously weighed more than one option. Managers did that in under 20 percent of cases. When they did, the success rate rose from 56 to 70 percent.
Considering an alternative is cheap on a spreadsheet purchase and expensive on an office opening, which is exactly backwards from how the effort usually lands. A big bet deserves the alternatives written out, the assumptions written as conditions somebody can check, the team's honest read before the CEO's is known, a premortem, and a date to look back. That is a few hours across a week. It is not a committee.
Why the everyday calls need five minutes
The other trap is the mirror image. A company that has been burned once starts running every decision as if it were a big bet, and the leadership team becomes a bottleneck through which nothing passes quickly. McKinsey's 2019 survey of more than 1,200 managers found they spent 37 percent of their time making decisions and judged more than half of that time wasted. Most of the waste is heavy process on light decisions.
An everyday call still deserves to be written down, in one line: what we decided, who decided it, why. Not for control. For memory. Six months on, the question "why do we use that vendor?" has an answer that is not "I think somebody chose it". Five minutes, one line, the person closest to the problem decides, and the leadership team never sees it unless it fails.
The sort in practice
When a decision arrives, before anyone opens a document, ask the two questions out loud.
If we are wrong, what does it cost? Put a number on it, even a rough one. A vendor switch might cost a month of somebody's time. A second office might cost a year of losses and the team you sent there.
How hard is it to undo? A subscription cancels. A lease does not. A hire can be reversed, at a real human cost. A public price change can be reversed, at a reputational one.
Costly and hard to undo: big bet, full treatment, the CEO's name on it. Anything else: five minutes, one line, delegate it to the person who owns the area and move on. The sort itself takes half a minute and saves the three meetings on the expense software.
Where Midfire fits
Midfire opens every decision with the same two questions and sizes the process from the answers: an everyday call is written in a few minutes and committed, and a big bet gets the paths, the conditions, the team's independent read and the look-back. The sorting idea is Bezos's and the evidence is Nutt's; the product's contribution is that the sort happens every time rather than when somebody remembers.
The decision my friend actually had
The office in the second city, by the way, was a big bet that got forty minutes because nobody wanted to be the one to slow it down. He ran it again properly: two alternatives, six conditions, four people's independent read, one premortem. It took a week. Three of the conditions turned out to be things nobody had checked, and one of them ended the idea. The expense software took him ten minutes on a Tuesday.