Deciding Under Uncertainty: When Nobody Knows the Answer

Why are the most important decisions made without sufficient data?

Should you hire a second salesperson even though orders fluctuate? Raise prices while a competitor is cutting them? Enter a new region for which there is no track record? Your reporting has no answer to questions like these, because it only shows the past.

Many owners respond with one of two patterns. Some postpone the decision until more information is available and find months later that the information has not improved. Others decide on instinct and cannot say afterwards what the decision was based on.

There is a third way between the two. It does not require better data; it requires a better procedure. The guiding question of this article is therefore: how do you make good decisions when the data are not sufficient? To answer it, we first clarify the difference between risk and uncertainty and then present three tools: the Door Grid, the process check, and the Three-Sentence Note. At the end we look at how your team can decide without everything landing on your desk.

Risk or uncertainty: what is the difference?

In everyday speech the two words are used interchangeably. Economics has separated them since the economist Frank Knight described the distinction in his 1921 book "Risk, Uncertainty and Profit." With risk you know the possible outcomes and can estimate their probability. With uncertainty you know neither all the outcomes nor their probability.

The distinction is practical because it determines the tool. You can estimate a payment default from your own numbers. You cannot calculate whether a new offer will be accepted. A spreadsheet with three scenarios then only creates the appearance of precision. Under uncertainty, small trials help more than big plans.

Reversible or irreversible: how much care does a decision need?

Amazon founder Jeff Bezos distinguished two kinds of decisions in his letter to shareholders for 2015. Some are consequential and almost impossible to undo; he calls them one-way doors. Others can be changed; they are two-way doors. In his words, the first kind calls for care and consultation, while the second should be made quickly by individuals or small groups with good judgment.

His point applies to small businesses just as well: most everyday decisions belong to the second kind and are still treated like the first. That costs speed without adding safety.

The Door Grid

In practice a second question is worth asking alongside reversibility: how much is at stake? Together the two questions produce a grid with four fields. It determines who decides and how much time the decision may take.

What does it cost not to decide?

Waiting feels cautious because nobody sends an invoice for it. The costs arise anyway. They just do not appear in any report, because lost orders are never booked.

A worked example: a business is considering hiring a second salesperson. The owner estimates that inquiries currently left unanswered would bring about 6,000 euros in contribution margin per month. Contribution margin is what remains of revenue after deducting directly attributable costs. If the owner hires someone and it does not work out, the attempt costs about 15,000 euros after a three-month probation period.

In this example, waiting costs more after three months than the worst outcome of the trial. The numbers are estimates, and that is fine. Even a rough estimate shows the order of magnitude of inaction. If you never write this number down, you compare a visible expense with an invisible one and therefore regularly decide to wait.

Waiting for certainty is also a decision. It just leaves the deciding to the calendar.

How do you recognize a good decision?

Winning a risky bet does not mean you decided well. Losing after careful deliberation does not mean you decided badly. In her book "Thinking in Bets," the author and former poker player Annie Duke calls the mistake of judging a decision solely by its outcome resulting. The term comes from poker, where good moves regularly lose and bad ones win.

The most dangerous field is the bottom left. An owner who accepts a large order without checking it and gets lucky learns that checks are unnecessary. Entrepreneurs who measure their decisions only by the outcome therefore learn the wrong thing.

The process check

A more useful question is: knowing what I knew then, would I decide the same way again? Four points make it testable:

• Did I compare at least two real alternatives, including "do nothing"?
• Did I write down what I expected before the outcome was known?
• Did someone who knows the matter disagree, and did I listen?
• Did I know beforehand at which signal I would stop?

The Three-Sentence Note: what you write down before deciding

The process check only works if there is a record from before the decision. Without one, hindsight bias sets in: once people know the outcome, they consider it more predictable than it was. If you noted nothing, you will later remember having suspected it all along.

The Three-Sentence Note prevents that with little effort. It consists of three sentences that you write before every larger decision and file with a date.

1. Expectation

   What do I expect, in numbers and with a deadline? Example: "The second salesperson brings in at least six orders a month by the end of March."

2. Reasoning

   Why do I believe that? Example: "About fourteen inquiries a month currently go without a fast reply."

3. Mind-changer

   Which observation would change my mind? Example: "If fewer than three orders a month come in after ten weeks, I end the trial."

4. Follow-up

   Set a date on which you read the note and compare expectation with reality.

The third sentence is the most important. It sets the stopping point while your judgment is still sober. In the middle of the trial, with money and effort already invested, stopping is harder.

After a year you have ten or twenty such notes. They show you where your judgment is reliable and where you regularly estimate too optimistically. One possible pattern: you set revenue too high and time frames too short. Knowing that improves every further decision.

How does your team decide without everything landing on your desk?

The Door Grid is also a delegation tool. When every question ends up with you, the cause is rarely a lack of courage among employees. Usually it is unclear what they are allowed to decide themselves. The grid answers that question in advance.

For sales this means, for example: your salesperson decides independently within a price range, and you see the numbers once a week. Only orders above a certain size or with unusual terms come to you. That way inquiries get answered faster and you keep the cases in which a mistake would be expensive.

What you can put into practice right away

You need no software and no workshop for this. One hour and a sheet of paper are enough.

1. Write down the three decisions you have been putting off the longest.
2. Place each in the Door Grid: reversible or not, low or high stakes.
3. Roughly estimate for each what another month of waiting costs.
4. Make all reversible decisions by Friday and write the Three-Sentence Note for each.
5. For the irreversible ones, check whether they can be set up as a trial; otherwise set a fixed date for the decision.
6. Create a folder called "Decisions" and enter the follow-up dates in your calendar.

You will probably find that two of the three decisions are reversible. Then you will have made them by the end of the week and have a clear head for the third.

Back to the opening question, which was how to make good decisions when the data are not sufficient. The answer lies in the procedure, not in better data: you make reversible decisions promptly and irreversible ones with more care, and you treat waiting as a decision of its own with a price. Whether a decision was good is judged by the process and not the outcome, and the Three-Sentence Note makes that process verifiable later.