The few decisions you can’t undo

Most doors swing both ways

In his 2015 letter to Amazon shareholders, Jeff Bezos split decisions into two kinds. Some are consequential and nearly impossible to reverse (one-way doors). They deserve slow, careful deliberation. Most are reversible (two-way doors): if you get them wrong, you walk back through the door and try again. Those should be made quickly, by high-judgment individuals or small groups. His warning was about treating reversible decisions with the gravity of irreversible ones. If you do, you get slowness, risk aversion, too few experiments and less invention.

That warning was written for a large company, where the usual failure is too much process. Early and growth stage companies tend to fail the other way. There is little process, and almost everything is decided quickly because there is no time to do otherwise. Fast is right for most of it. The risk is that a one-way door gets walked through at two-way speed, because it looked like every other item on the list that day.

Where the one-way doors hide in consumer health

In consumer health, some of the least reversible decisions do not look important when they are made.

Words. The FDA decides whether a product is a regulated medical device based on its intended use, and it judges intended use objectively: from labeling, advertising and other statements the company makes, and from the product’s design. The agency’s general wellness guidance, updated in January 2026, applies the same test: a product whose labeling, advertising, user interface or functionality refers to a specific disease or claims clinical-grade accuracy is not a wellness product. The same sensor can fall under the FDA’s low-risk wellness policy or under device rules, depending largely on what the company says it does.

Study design. The first study a company runs often defines what it can claim for years. Endpoints chosen for speed or cost become the evidence base the brand has to live with.

Money and people. The first financing terms and the first senior hires set expectations that are expensive to unwind.

A founder’s early decisions and the conditions at founding shape a venture’s growth and survivability well past the founding phase. One recent study in Cornell’s Administrative Science Quarterly traced how a founder’s early choices at the optics maker Zeiss became structural, behavioral and product features that lasted more than a century. That is one company and an extreme time frame, but the mechanism is familiar: what gets decided early gets built, and what gets built early is hard to change later.

The cost of keeping the door open

Walking through a one-way door too quickly is one mistake. There is an opposite mistake. If a few decisions deserve more deliberation, many deserve less, and the cost of leaving them open is higher than it looks.

In a 2002 study, psychologists Daniel Gilbert and Jane Ebert gave photography students a choice between two of their own prints. Some could swap their choice later; others could not. Most people said they would rather keep the option to change their minds. The students who had that option ended up liking their choice less.

A 2011 follow-up suggested why. While a decision stays reversible, people keep thinking about it. Open decisions reduced working memory capacity and increased regret.

In our experience, a decision that is never quite closed keeps coming back: in the next meeting, in the late-night message, in the second-guessing that follows every setback. For a CEO already juggling many decisions, closing the reversible ones is not just efficient. It frees attention for the few that matter more.

A simple sort

Three questions separate most one-way doors from the rest.

1. What would it cost to undo this in a year? Not in money alone: in regulatory standing, credibility with investors, trust with customers, and the team’s time.
2. Who outside the company will rely on it? A decision that a regulator, an investor or a retail partner will build on is harder to reverse than one that stays inside the building.
3. Can it be made smaller? A pilot, a phased commitment, a limited launch or an exit clause can turn a one-way door into a two-way one.

A decision that is cheap to undo, relied on by no one outside, or easy to shrink should be made quickly and closed. A decision that fails all three deserves slower, more deliberate treatment and usually the CEO’s direct attention.

Recognizing that in the moment is most of the work. The rest is making sure the few decisions that deserve slow thinking get it, and that everything else gets decided and left alone.