Biases in decision making

Controlling risks for startups -3

In the previous articles, I started a discussion on ways to reduce the impact of ‘luck’ in the very early stages of a company. I first discussed the method of ‘rapid skill improvement’ and the second method of ‘controlling risk exposure’.  This week I discuss the third method of cutting out biases from decision making.

In my previous article, I explained that biases stunt our learning curve. They don’t allow us to draw the right lessons from our experiences. Not learning lessons quickly enough reduces our rate of skill growth and this can be fatal to early stage companies who don’t grow fast enough before they run out of money. We said that failures and success create learning loops (as shown below) – and biases lead to incomplete learning loops and hence a flat learning curve.

So far, I’ve covered 4 different biases – survivor bias or outcome bias, where we overestimate the probability of success and underestimate probability of failure. Self serving bias is where we overprice our own skill and under-price the role of luck. A similar bias that fuels self-serving bias is a confirmation bias – where we actively seek out only the information that is consistent with what we believe. Envy and authority bias is when we under-price the skill of others and overprice the skill of others respectively.

Other important biases that we haven’t explored are ostrich bias – where we don’t change our path or position even in the presence of a lot of contradictory information. We deliberately ignore and act as if that information does not exist. This is sometimes also referred to as commitment bias.


Anchoring bias is to arbitrarily anchor yourself on a position or a number just because you heard a random piece of information. To study full universe of biases, you can refer to a talk by Charlie Munger on ‘the psychology of human misjudgement’ – Charlie has been a successful investor for decades and his summary on cognitive biases is a must read for everyone active in decision making.

Some common examples of these biases in the startup world:

Technology will make XYZ obsolete in 10 years..

Example of confirmation/ostrich and/or self-serving bias

You take an absolute, definitive position about the future. There is a high chance that a self-serving or confirmation or ostrich bias is at play. Look out for words such as ‘must, absolute, sure’ in your statements – these biases usually are hidden behind such words.

“I will be Airbnb of this or Amazon of that”..

Example of survivor bias

Another thing we usually hear is While this statement is usually made to make people understand the business quickly, I find such statements loaded with survivor bias. For every Amazon, there are a million startup that tried becoming an Amazon of a niche market and failed. Trivializing a business by benchmarking to a biggest incumbent can make one overestimate probabilities of success than what they actually are.

We should raise at $$$ because ABC raised at $$

EXAMPLE OF ANCHOR BIAS


Another common variant is where an entrepreneur or investor values the business to an arbitrary anchor. Anchoring bias is severe in startups that are highly illiquid where price discovery in open market is non-existent.

We need to hire a senior MBA as VP, Sales

AUTHORITY BIAS


One form of authority bias is when founders think some guy with a flashy MBA will come and drastically improve the sales – this makes founders either dislike the sales process or spend on an expensive sales guy early in the game.

How to cut down biases?

Now that we know these biases exist, lets get to some generic methods of cutting down biases from decision making. I will not go into each type of bias but give a generic framework to cut down bias in decision making

1. Seek diverse perspectives

You need 3 set of people in your life – whose perspectives actually matter.

  1. Your co-founders and investors – people who are truly invested in your journey and who have skin-in the game. I’ve always believed this as an entrepreneur that entrepreneurship is not worth doing without a co-founder. And this is precisely why – you need someone who you can deeply engage with to discuss things as they are.
  2. A set of people who deeply care about your success. These are people who are high on integrity and people you personally know and have observed for for a long time – it could be your spouse, your maternal uncle, your friend or a true mentor. Please be advised that in this world, advice is something everyone gives because its free (If you noticed, I just gave one). I’ve always stayed away from people who give advice in exchange of money – regardless of how accomplished someone is, advice for money is a bad deal
  3. People who are in the industry as you are but believe in the exactly opposite thing as you do. They could be victims of ‘confirmation’ and ‘self serving’ bias and they could go out of their way to find faults in what you are doing. This might help you unearth something that you never considered.

2. Purpose Filter

Second, all choices and decisions should be subservient to your purpose in life. A decision that is inconsistent with your purpose has greater odds of failure even if worked for others. I define my purpose filter as a 6 level filter – you can define something that has even more levels than I have.


At the top, I start with purpose, the fundamental question of ‘why I am doing something’ – second is your vision of ‘what you want to make’ in this world, third is your mission, a set of principles that guide you on how to achieve your vision. Fourth is your five year goal – what do you want to achieve in exactly 5 years time – 5 years is just about right, not too long, not too short. Fifth is your 1 year Objectives and Key metrics – popularly called OKR’s. And then you distil it down to what results you want to achieve in next 90 days. Overtime, when you pass all choices through these filters, your decisions will rid themselves of intrinsic biases

3. Time Travel


In this chart, I’ve marked a timeline, At the center is T=0, which is today, on the left is the past and on the right is the future. As on today, we are faced with a number of choices – but we choose one choice of the many, we call that our decision and move ahead. In future there could be a positive outcome I call (Op) and a negative outcome I call (On). 

When we move from our present into our past, we call that activity as a ‘post-mortem;. Mapping where we are today with the choices we had in the past & the decision we took in the past to get here. Analyzing whether that decision was the right bet at the time and asking if there was something you could have known but didn’t because of certain bias. In this exercise, you have the wisdom of time and experience to guide you. Purpose of these post mortems is to create processes and habits that guide you to better decision making in future.

Second tool is what we call a pre-mortem – assume a negative scenario (On) pans out in future and trace your steps back to the present. Freezing a negative outcome allows us to make logical connections better and identify useless decisions that we attach a lot of importance to. It forces our mind to analyze quality of decisions and that usually is bad news for hidden biases
Third tool is what we call ‘backcasting’ – visualizing a success and then moving backwards to analyze what are 2-3 big decisions that have the biggest impact on that success. This is exactly opposite to pre-mortem – by being on both sides of the curve, your mind is automatically more open-minded and becomes capable of objective analysis

Fourth tool is to map all possible scenarios in future – what if the 3rd wave of Covid causes another shutdown, what if I don’t end up winning a pilot, what if I lose my CTO, what if we don’t find capital in 3 months.. Mapping all possible scenarios is called ‘Reconnaissance’. Assigning probabilities to these futures is what I called Scenario Analysis – a probabilistic weighting of scenarios makes reconnaissance lot more powerful. Probabilities might go wrong but the analysis will be useful nevertheless.
Ultimately the effect of all these time travel tools is to make you more open-minded and shun ‘absolutes’ in thinking. Absolutes are usually outcomes of biases – by shunning absolutes, we remove the power of biases to control our decision making.

To summarize our key learnings of this episode,

  1. Biases stunt our learning curve and hence create risks
  2. Many biases exist – crucial ones are survivor/outcome bias, self-serving bias, ostrich and anchor bias
  3. Create a core group of people and actively seek out perspectives
  4. Pass every decision through your purpose filter. If its not consistent, drop it
  5. Use time travel tools that connect the past you with the future you.