Making decisions in uncertainty

When I think about the best leaders I’ve worked with, there has always been one element that stands true. I trust and respect their decision making. I don’t have to want to be their best friend, but I have to believe in the decisions they make.

I’m faced with a multitude of decisions every day, and my biggest priority, both professionally and personally, is to constantly improve how I make them. So I thought I’d share the four most important principles I’ve developed. I’ve added further reading within each one, and if you’re keen to learn more I’d highly recommend the podcast and blog of Farnam Street - it’s a goldmine of wisdom in this field.

I first wrote this in 2019, when I was at Weploy. Most of it has held up. Some of the science behind it hasn’t, so I’ve updated it - and been honest about what I’d no longer cite.

1. Prioritise the decisions you make

We’re faced with a paradox of our own making. People who lack choice want and fight for it, but choice has a cost. Every decision takes attention, and attention is finite.

When I first wrote this, I leaned on two famous pieces of research to make that point. One was the idea of “decision fatigue” - that willpower runs down like a battery the more you use it. The other was a study showing Israeli parole judges ruled more harshly later in each session, recovering after a food break.

I wouldn’t cite either with the same confidence now. In 2016, a replication run across 23 labs with around 2,100 participants found the “ego depletion” effect behind decision fatigue was close to zero. And the judges study has been picked apart: the order of cases wasn’t random, and Andreas Glöckner’s simulations suggested judges sensibly avoiding complex cases just before a break could explain much of the pattern.

So the battery metaphor is shaky. The conclusion still stands, for a simpler reason: you only have so much time and attention, and the trivial decisions - which coffee, which shampoo, whether to snooze the alarm - eat into it just the same.

So to improve, limit the number of decisions you make. A great tool I found on Farnam Street was a take on the Eisenhower matrix that sorts decisions into four categories:

1. Irreversible and inconsequential
2. Reversible and inconsequential
3. Reversible and consequential
4. Irreversible and consequential

Decisions that are inconsequential, whether reversible or not, should be delegated. Decisions that are consequential but reversible are the opportunity to run experiments, where your team can try things and share the learnings.

The only decisions you should focus on yourself are those that are consequential and irreversible.

The same idea has since been put more memorably by Jeff Bezos. In his 2015 letter to Amazon shareholders, he described irreversible decisions as one-way doors, to be made “methodically, carefully, slowly”. Most decisions are two-way doors - if you walk through and don’t like it, you can come back - and those should be made quickly by people with good judgement. His warning is the one I see most: as organisations grow, they start treating every decision like a one-way door, and everything slows down. The following year he added that most decisions should probably be made with around 70% of the information you wish you had.

I’ve found the trick is often to turn a one-way door into a two-way one. At Rescale I wanted to change how our BDRs were paid - moving most of their plan from opportunities created to meaningful interactions, with a smaller part on their pod’s revenue. A comp change feels about as irreversible as decisions get, and it was a hard sell: meaningful interactions sounded “fluffy” next to an opportunity count. So rather than argue it out as a one-off, all-or-nothing call, we ran it as a pilot. That made it reversible, and the results made the case far better than I could in a meeting.

Of course this is only a framework, and you have to define what consequential and reversible mean for you and your business. But it has helped me focus on what matters, and build capability and empowerment in my teams.

Further reading: an introduction to decision making, Farnam Street.

2. Give yourself time

It’s easy to think great decision makers are the ones who can instantly weigh up a scenario, see the possible outcomes and decide on the spot. It’s impressive. But what’s the rush?

If this really is a consequential, irreversible decision, the urge to decide immediately is usually a bias pushing us to do what’s easy with the information in front of us.

It’s hard to think of opposing points of view or other outcomes, but that’s really what success in my job is all about. So I try to apply as many lenses to a situation as I can.

Then I give myself thinking time.

The clearest recent example for me was a key hire in Japan. It was a critical role, it was hard to fill, and the business understandably wanted it filled fast. But a hire like that is about as close to a one-way door as leadership gets - get it wrong and you pay for it for a long time, and so does the team. So I resisted the pressure to grab the first good-enough candidate and gave the decision the time it deserved, even with the clock running.

For me that means changing my environment - going for a walk, or sitting in another part of the office - to let my mind wander around the decision. I play thought experiments and walk through the outcomes with different hats on.

One great model comes from the mathematician Carl Gustav Jacob Jacobi, and was made famous in business by Charlie Munger: invert the problem. Rather than trying to solve a problem forwards - say, increasing productivity in your team - ask what are all the things I could do to reduce productivity in my team. Then don’t do them! It may not solve the problem, but it stops you building in the arguments that lead to poor outcomes.

There’s a team version of this worth adding: Gary Klein’s premortem. Before a big decision, ask everyone to imagine it’s a year from now and the decision has failed, then write down why. It gives people permission to raise the doubts they’d otherwise keep quiet. (The research behind it found imagining failure helped people come up with around 30% more reasons - more reasons, not necessarily better ones - so treat it as a way to widen the conversation, not a crystal ball.)

AI can make a useful sparring partner here too - ask it to argue the other side, or to run the premortem with you. It’s good at widening the list. It’s also very good at sounding certain, which brings me to the next principle.

Further reading: avoiding stupidity is easier than seeking brilliance, Farnam Street.

3. Be accepting of ignorance

First, let’s define what a decision in uncertainty means.

The distinction goes back to the economist Frank Knight, who in 1921 separated risk, which can be measured, from uncertainty, which can’t. In a decision under risk you don’t know the outcome, but you do know the probabilities. A professional poker player knows the odds of winning with their hand, even though the result is still unknown. In a decision under uncertainty you don’t know the outcomes or their probabilities. Knight argued that’s exactly where judgement - and profit - comes from.

So when you’re dealing with uncertainty, you will be ignorant of some things that matter. But importantly, there are two states of ignorance:

1. Primary ignorance: if you can’t see it or understand it, it doesn’t exist.
2. Recognised ignorance: you recognise you can’t predict all the outcomes.

It might seem that if we can’t predict the unknowable any better either way, why bother? But when you’re aware of your own ignorance, you can start looking for the biases that could cloud your decision. There are plenty, but the three I find most relevant are:

1. Overconfidence bias: our confidence in our own beliefs is higher than their actual accuracy.
2. Salience bias: we focus on information that is prominent and easy to see, and ignore what’s less so.
3. Confirmation bias: we recall examples that reinforce what we already believe.

The biggest addition since 2019 is that bias isn’t the only problem. In Noise, Daniel Kahneman, Olivier Sibony and Cass Sunstein showed how much experts disagree with each other on the very same case. When they asked senior executives how much two experts’ judgements might differ, the typical answer was around 10%. When one insurer audited its underwriters pricing identical cases, the median difference was 55%.

Bias is when everyone is wrong in the same direction. Noise is when everyone is wrong differently. That’s a strong argument for getting several independent views on a big decision before people discuss it, so the loudest voice doesn’t become the only one.

So I try to embrace my own ignorance, and to uncover how my decisions - and my team’s - could be coloured.

Further reading: the anatomy of ignorance, Devjani Roy and Richard Zeckhauser.

4. Optimise for the long term

Another factor I find important is optimising decisions for the long term. Again, we’re hard-wired to seek instant gratification.

For me this comes down to looking for win-win outcomes and what’s called second-order thinking. They’re strikingly simple in theory but take conscious effort to put into practice.

Any relationship, if it’s to last, has to be win-win. So when I’m considering which direction to go, it has to benefit both the customer and the business if it’s to have any chance of prospering over time.

Second-order thinking can be boiled down to a simple phrase.

And then what?

If I make this decision now, what happens tomorrow? In a week? A month, a year, ten years?

Not every decision needs you to look that far ahead, but thinking through the branches of possible outcomes is hugely valuable. I regularly find decisions that are negative in the first order - a high cost, a slowdown - turn out to be net positive in the second and third.

And as my decisions should build long-term, sustainable advantage, second-order thinking is essential if I’m to succeed.

Further reading: second-order thinking, on Medium.

---

So those are the four principles I try to use, updated with what I’ve learned since I first wrote them down. I hope you find them valuable, and I’d love to hear the tricks and models other people use to improve their decision making - especially any you’ve had to unlearn.