How to make good decisions
Judgement is the most important skill
As we enter an age of AI native startups, it’s easier than ever to build, which only increases the importance of good judgement and decisions.
Most startup advice focuses on speed of execution, but much less on what decisions matter, and how to make them well.
What decisions are important?
At Prolific, only a few important events and decisions really made the difference to success.
Resources, capital, talent, and risk should not be allocated evenly in startups, but concentrated where they can create exponential impact. The tricky bit is to see which decisions might have those asymmetric consequences, and make them well.
I’m a fan of this framework from Patrick Collison of Stripe.
The decisions which are important to consider carefully are
High Magnitude/Hard to Reverse
Easy to Reverse/High Magnitude
High Magnitude/Hard to Reverse decisions require the most robust decision making process.
Easy to Reverse/High Magnitude require a good level of confidence, and importantly a process for tracking and adjusting.
It’s also easy to miss whether decisions are reversible or not. Some choices feel like big commitments, but could be reversed with some friction, upset, or perhaps embarrassment. For example a building lease, supplier contract, or setting up a new team.
Some decisions feel trivial, but could become critically embedded, with 2nd order effects, such as technology choices for an MVP, or work from home guidance.
How to build an autonomous decision making culture
The decision making culture at your startup will influence this, who gets to make specific decisions, how democratic are they?
“No Rules Rules” which deals with the culture at Netflix, emphasizes giving employees the freedom to make decisions, coupled with the responsibility to act in the company's best interests
However this doesn’t really just mean there being "No Rules". This is where in my opinion, the book gets misinterpreted.
At Netflix there are certainly “cultural norms” - you are expected to optimise for certain outcomes, and to make decisions in certain ways - aren’t these implicit rules? If you don’t you’ll get a reminder of the “non-rules” through feedback, and performance review. Looser rules can only go in hand with strong accountability for results. At Netflix you’ll be let go for poor results of your decisions, a feature which too many leaders especially in the UK struggle with.
In building a decisions making culture, consider what kinds of decisions you are truly incentivising (re risk taking). Do you really reward your team taking smart risks and failing, or for playing it safe?
When to make a decision - Hunter S Thompson v Angela Merkel
There are 2 ways to think about delaying decisions:
“A man who procrastinates in his choosing will inevitably have his choice made for him by circumstance.” Hunter S Thompson.
Angela Merkel, the former German Chancellor, was known for delaying important choices until the last possible moment - "merkeln"
Kicking the can down the road can be effective, because a lot of problems will go away by themselves. Sometimes paths become obvious, or options reduce, making contentious decisions politically easier. But this is really a form of deciding which decisions are actually high impact, and which don’t matter. It can help with politics, but this shouldn't be your main concern.
Optionality should be maintained only when it’s valuable , and decisions should be made as soon as it’s possible to make them confidently - there should always be a bias towards speed and clear planning in startups.
Not deciding is often deciding because inaction will have consequences. Often apparent when not addressing cultural or performance issues.
Being “data driven”
I feel like leaders claiming to want to always be "data driven" covers up for lacking a rigorous framework and mental models, for how to make good decisions, and when data matters.
Early only Founders can often work around this due to their intuitive feel for what customers want. As you scale, you need to figure out what decisions matter (for many too much time is wasted on them - the bike shed fallacy), and how to make them.
People also often underate the opportunity cost of getting useful data, it might be the best way to make a decision, but waiting might not be worth it. Particularly early on, when your user numbers are tiny.
Wittgenstein's Ruler, a concept introduced by Taleb, states that when measuring something, you may actually be learning more about the measuring tool than the object being measured.
"Unless you have confidence in the ruler's reliability, if you use a ruler to measure a table you may also be using the table to measure the ruler" - Nassim Nicholas Taleb (Fooled by Randomness)
Startup metrics and attribution models can act as unreliable rulers. When measuring business performance, the data might tell you more about your measurement biases than your actual business growth, so be wary of taking data at face value when making decisions.
If you want to get good at decisions and prediction you need to practice. Texas road gamblers would hone their skills by constantly making wagers and predictions on everything they saw, such as when is the next train coming, which fly will crawl up a window. You can do the same, and as we saw in the “teaching a horse to sing example” it’s useful to log key decisions, predictions, and outcomes in a decision log. To learn, for transparency, and as a tool in building an aligned decision making culture around risk, ethics and goals.
Methods and useful models for making decisions
In Principles Ray Dalio explores some interesting ways to formalise decision making frameworks in organisations.
For example "Believability-Weighted Decision-Making" suggests giving more weight to advice from people with proven track records in the specific area of decision-making. This approach helps filter out less reliable opinions and focuses on expertise.
You might find it impractical to implement such structured methods, but some key elements like controlling emotions, root cause analysis, weighting advice, measuring results, surfacing principles, long term visualisation, 2nd and 3rd order effect, and most importantly tracking results should be incorporated.
There are many useful mental models that can help with making decisions. Charlie Munger’s “Poor Charlies’ almanack” is a great exploration of some of these, and the biases we are prone to.
His "lollapalooza effect" concept (when several factors or forces work together in the same direction, creating a result that is more powerful than the sum of its individual parts) has become a favourite of mine in trying to explain, or cause large effects - and due to the power law, large effects are what you are looking for in startups.
There are many useful models highlighted here in particular I like to use first principles thinking, and inversions. What's your favourite?
Rules, razors, and heuristics can be helpful, but you need to be careful about leaning on them as a substitute for really thinking. I think they are best aligned to operational principles you can use in your team. For example you might have "if in doubt go with the fastest, least complex option" or whatever is appropriate to your team culture.
