Product Management from First Principles

I recently quoted the proverb To learn is to read, to understand is to write, and to master is to teach. It has since occurred to me, I have not been doing enough of the latter recently. So to master my own craft and share my learnings and understandings of product management with others, I have started Science Not Ego, which I hope will become a useful resource on the science (and sometimes art) of product leadership. This phrase has become an internal mantra in following the scientific and outcome-driven approach I use to lead teams to build great products, I hope it will do the same for you.

When deciding what topic to start with, I went back to first principles…quite literally…the way I begin to tackle most complex problems.

First principles by definition are:

The fundamental concepts or assumptions on which a theory, system, or method is based

First principles are a key part of a product manager’s toolkit when tackling complex problems, where there may be many variables or assumptions in place. When used effectively they allow you to scientifically solve problems and create solutions beyond what would be possible when using a comparative approach based on convention, analogy or previous experiences (notably all things which fuel a person’s ego).

I’ll give an overview of where the methodology comes from, some personal experiences and 3 easy steps on how you can harness this method when designing solutions to problems.

Where are first principles used?

The methodology is heavily used in applied sciences such as physics or engineering and is referenced as early as Aristotle in Physics, and Descartes in his Method of Doubt. In more modern times, Henry Ford or Elon Musk famously have adopted first principles thinking. Put simply if it weren’t for this approach we’d have settled for faster horses or had to wait another 20 years for affordable electric cars or reusable rockets. While others wrote off the manufacture of cars for the masses, Ford asked:

“What would it cost to build a car if I broke it down to its most basic materials and found a more efficient way to put it together?”

Musk has since done the same with both Tesla and Space X overcoming the seemingly prohibitive cost of battery and rocket manufacture.

Elon Musk and Henry Ford — Product pioneers who make use of first principles
“I think generally people’s thinking process is too bound by convention or analogy to prior experiences. They’ll say, “We’ll do that because it’s always been done that way.”
“[With first principles] you boil things down to the most fundamental truths…and then reason up from there.”
— Elon Musk

Personally, I had the methodology impressed upon me by my Father from an early age who taught me the etymology of new words and axioms of science as I learned new concepts. More recently during my engineering degree, my tutor famously set 2 hr exams, with a single question, around complex systems where it was required to derive all answers from memorised first principle equations. If you had not mastered this technique there was no chance of you being able to ‘calculate the flow rate of a viscous fluid in a pipe between two vessels of irregular geometry as shown in the diagram below’ — I may have that experience etched into memory!

Etymology is a form of first principles thinking — this conversation is very similar to many I had with my father

How do they apply to product management?

In a more practical product management example I used the first principles approach when tasked with leading a team in designing a product that would improve the energy efficiency of a household washing machine. Starting from first principles, understanding that the majority of energy was used during the spin cycle and that the most energy was expended in driving the rotational force of the drum, we started here. We found that due to the fixed drum size and weight the rotational energy was consistently high, relative to the weight of laundry — completely inefficient for smaller loads. So the resultant design was based on a variable size washing drum which optimised the drum size to the load and therefore decreased the energy required, per weight of laundry, to force the water out during the spin cycle…remember spinning a ball on a string of various lengths during school physics lessons?

3 steps to working from first principles

This all seems a bit theoretical so here are some simple steps to add first principles thinking to your product tool box:

1. Understand the current state and the desired state AKA Vision

When solving a problem or creating a product the first step is to have a good understanding of the current situation for customers, what they are currently doing and the problems they are facing. Basically what reality currently looks like to them.

Then work out what the customer is trying to achieve i.e. what their desired state looks like. Jobs to be done is a good technique for getting to the bottom of this.

The desired outcome or vision should be clear at this point otherwise you will miss the target on the next steps.

2. Peeling back the onion or 5 x Why? AKA First principles

Typically a product manager may look at this information and try to solve the bottlenecks based on convention or analogous data. This will only tell you what has happened in the past, not what is possible in the future. Now’s the time to step back and apply first principles.

The first step in getting to the route of an issue is finding out what information we are confident is an axiom or universal truth. Ask “are we sure that is true?” until the answer is yes.

Either you can start from one of these truths you are confident applies to the problem — inside out. Or if you are unfamiliar with the area a good technique is peeling back the onion to get to the one thing you know to be true or using the 5 Whys technique — outside in. By breaking down each assumption or variable to the point where you are left one or two ‘truths’ you are now in a good place to start building up a hypothesis to test.

3. The scientific method AKA strategy

Now it is time to build up the product or solution based on a series of assumptions or hypotheses which need to be tested to the extreme and adjust based on learnings until you reach your desired outcome. This is a typical Lean Methodology.

Be aware that as you start to prove, or indeed disprove hypotheses, you may need to adjust your thinking on related assumptions as you go. To be most efficient with this I like to start with the assumptions which are pivotal to the argument, that way you can build up the theory on a strong foundation. Testing a simple theory first may knock out the whole hypothesis and you may have to start again with wasted effort. Riskiest Assumption Tests are a good technique to follow here.

A practical product example

To show how this could work in a practical product example here are how these steps may have been used when designing Klarna’s Buy Now Pay Later product, which allows customers to receive credit on the fly in an eCommerce checkout, with just customer email and address details.

i) Understand the current state and the desired state AKA Vision

Current state: It is difficult for customers to apply for credit online as it requires many pieces of personal information to run a credit check on a customer

Desired state: Applying for credit online is as simple is as smooth as paying on a 1 Click checkout

ii) Going back to first principles

Q1: Why are multiple pieces of information needed to run a credit check?
A1: So you can understand a customer’s payment history, amount owed, length of history, types of credit
Q2: Why do you need to know this information?
A2: So you can calculate the risk to the bank in lending the customer credit
Q3: Why do you need to calculate risk?
A3: So the bank can work out how much money is safe to lend

Therefore we can be sure: Credit rating is a function of risk

iii) The scientific method — testing hypotheses

So now we know credit rating is a function of risk so how can we measure the risk of lending just by knowing a customer’s basic details e.g. name, address and email?

Test: build a risk algorithm to determine if customers will pay or not using minimal customer information and compare the outputs to traditional credit scoring. Iterate until the outputs are as good or indeed better!

It is likely that this is the sort approach Klarna took when creating this product and has resulted in them being well ahead of traditional credit lenders who still operate on the paradigm that long forms of customer information are needed to validate their credit score.

Tying it all together

Establish the status quo and the desired outcome, strip back external biases to what you know to be true, build up a set of hypothesis and test them in order of importance until the theory is watertight — well done you have applied first principles thinking to product management!