5 min read
Next in trending

Making Decisions Under Uncertainty

Or, How to Decide with Incomplete Data

Making Decisions Under Uncertainty

Or, How to Decide with Incomplete Data


“In any moment of decision, the best thing you can do is the right thing. The worst thing you can do is nothing.” — Theodore Roosevelt

As product managers, entrepreneurs, and leaders, we are often confronted with the challenge of making decisions under uncertainty. This is especially true in situations where we are trying to decide something based on the external environment—market trends, customer needs, or competitor reactions. To be an effective leader, however, it’s important to be able to make the right decisions in a timely manner, despite the uncertainty.

During my time at McKinsey, we were often called on to advise a client to make an important decision without the benefit of a lot of data. A good example was when a client asked us to evaluate whether it should move into an adjacent, but new, market. We often didn’t know how that market would grow over time, or what kind of market share our client would get in the new market.

As McKinsey consultants, we were trained in how to make decisions and recommendations in just these types of situations. Below are some of the tools we used to make decisions with incomplete data.

1. Day One Hypothesis

At the beginning of any new client engagement, we were expected to develop a “Day One Hypothesis.” Based on the high-level facts that we had learned within the first 24 hours of the project, we were forced to develop an early hypothesis of what the solution to the client’s problem was.

How you develop your hypothesis is a combination of good problem solving skills, pattern matching, and intuition. We were also often encouraged to “get smart” about a particular area by reading 3rd-party analyst research, or interviewing industry experts. But the expectation was that within your first day on the project, you should come up with your early hypothesis about the right decision to make. Then, you can always gather additional data to refine your hypothesis over time. You should be open to changing your hypothesis as you gather additional data, but you should start out with a hypothesis on Day One.

The reason why this approach is helpful is that after Day One, you always have a decision that you can stand behind at any point in time. You may try to gather additional data, but based on the data that you have already reviewed at that point in time, your current hypothesis is your current decision. This approach works best when you don’t know how much time you have to make a decision, but you could be called on at any time to make it.

2. Directionally Right, Same Order of Magnitude.

A lot of times, people get hung up on making a decision because they want to make sure they get the right answer. This is especially true when you are trying to get to some numerical value: If we decide to do this, how much revenue can we expect in year 1, year 2, and year 3 after we do it?

At McKinsey, one of the things we were taught is that your answer needs to be directionally right, and the same order of magnitude, as the “right answer” or actual value. Without having complete data, you will never be able to get to a precise value. Instead, you should strive to make sure your answer is directionally right. Will we be revenue positive, neutral, or negative if we make this decision? And you should strive for getting the order of magnitude right. Will we increase revenue by $1m or $10m if we do this?

By approaching the decision this way, it takes the pressure off to get to the precise right answer and avoids “analysis paralysis.” You need to be right enough with your decision, but you don’t need to be perfect. If your decision made the company $5.2m or $4.8m net profit (instead of the +$5m you estimated), chances are that the company would still look at the decision as being a success. However, if your decision cost the company $10m in profits (instead of the +$5m you estimated), then your answer was neither directionally right, nor the right order of magnitude—and your decision would be a failure.

Without having all of the data—and you will never have all the data—you can’t be expected to forecast future value with a lot of precision. So strive to be directionally right, and the same order of magnitude.

3. What Do You Have to Believe?

The final tool we used for making decisions with incomplete data was an exercise called, “What do you have to believe?” This was a useful decision-making tool when there was some driver of our decision which was very uncertain. Suppose we were trying to decide whether to enter a new market, and one of the key drivers of our revenue estimate was what our market share would be in three years. We would come up with various scenarios that would assume different levels of market share, and then calculate what the potential revenue size would be.

To make the example more concrete, let’s assume that in order to get the green light from senior management to pursue a new market opportunity, we need to be able to show that we will get to $50m revenue in 3 years. We are fairly confident in our projection that the market size will be $250m in 3 years. Currently, there are a handful of other competitors in the market, each with roughly 10-20% market share. Sitting here today, we have no idea what our market share would be in this new market in 3 years.

So we would ask ourselves: “What do you have to believe for this to be a $50m opportunity in 3 years?” The key assumption you have to believe is that we would achieve 20% market share in 3 years. Given the current competitive structure, trying to achieve 20% market share is within the realm of possibility. And perhaps we have shown that we can enter new markets in the past, and have the experience to gain meaningful market share when we do so. Then we would have more confidence that we actually could achieve 20% share.

What if the circumstances were different: there were two big incumbents in the market (50% and 40% share respectively), and a bunch of smaller players each with 1-2% share. And if this was an entirely new market/business model for us to enter? Then our confidence in our ability to achieve 20% share would be much lower.


One of the hallmarks of being a successful product manager, entrepreneur, and leader is the ability to make decisions. In our fast-moving world, we will often be called upon to make decisions with incomplete data. Rather than falling into the “analysis paralysis” trap, where we spend too much time researching and collecting data, we need to be able to deal with the ambiguity and uncertainty and make the best decision we can, with the data that is available to us at the time.

At McKinsey, we were taught three approaches to making decisions under uncertainty:

  1. Day One Hypothesis
  2. Directionally Right, Same Order of Magnitude
  3. What Do You Have to Believe?

These three tools have been immensely helpful to me in my own career, especially when I’m asked to make a decision about a future business opportunity where there are some key unknowns. I’m confident that these approaches will help you make decisions under uncertainty, and make sure that you don’t end up in the worst situation, as Teddy Roosevelt said—doing nothing.