Wisdom from Patrick Collison of Stripe on the Farnam St podcast

I absolutely loved Shane Parrish’s conversation with Patrick Collison from Stripe, so I thought I’d re-listen and take notes this time…

How would you describe the culture at Stripe?

Prizing rigour and clarity of thought — not smoothness and unruffled feathers. Prefer correctness over cohesion. Want people who don’t mind being wrong and are open to new ideas.

Determination and competitiveness — doing anything of significance is hard, so they need to want a hard challenge and be fulfilled by that. Be able to cope with ambiguity and things that are broken.

Interpersonal warmth and a desire to make people around them better off— they are simply nice to other people.

What have you learned from scaling Stripe?

It’s both relatively straightforward and extremely hard.

It’s straightforward because you can usually see the problems that need to be solved, and they usually aren’t outlandishly difficult to solve (hiring someone, raising capital etc).

It’s extremely difficult because you can’t control time, so the problems appear much quicker than you can solve all of them, even if each one of them is reasonably manageable.

That becomes emotionally quite difficult to manage as more and more challenges appear every day — it seems so unreasonable — you have to self-manage.

There’s also the problem of dealing with uncertainty. You are often making decisions that are difficult to course-correct from in the face of uncertainty. The uncertainty isn’t necessary because with time and research you could find out the correct decision, however you simply don’t have the time to lessen the uncertainty.

Constantly trying to balance exploration and exploitation. How to do what you’re already doing well, and also exploring what else you could be doing. Optimising outside the system vs optimising inside the system.

The personal challenge of having to learn the skills of operational management and leadership on a timeline that’s out of your control, and this can be very hard. You just have to try and learn these skills as quickly as possible.

How do you know whether to go broad or deep in terms of focus?

We invest most of our effort, 70–80%, on optimising what we already have, and roughly 20% into things that are more speculative.

The question is what should those percentages be? And should all teams have the same percentages or should those differ? You need to continually test this to ensure you avoid the diseconomies of scale that otherwise set in as you grow.

How do you know which ideas to explore?

We considers all the factors (effort, chance of failure etc) and choose the one with the best returns profile.

For example, when deciding whether to launch Atlas we thought: if it didn’t work it wasn’t going to create that much downside for Stripe, it wasn’t going to require large fixed cost to find out if it’s initially working, if it did work the returns could be significant, the kinds of things we had to do for it would be things that are valuable in other parts of the business, so we’d learn new capabilities and skills etc etc.

You think about making these decisions like you think about making bets.

Buxton index: large organisations have longer time frames to make decisions in. Organisations with different Buxton indices find it very difficult to work together.

How has your ability to place massive bets changed as the company has got bigger?

Now we have to make bets that don’t have massive/fatal downside, or fatal downside across a portfolio of bets.

The challenge is to reconcile the mindsets of different parts of the company, when some teams are trying to do something fundamentally different to what already exists, and others are trying to do what they’re currently doing better and better. You need to have both sets of people, and have them at the end of the day have dinner together while fundamentally disagreeing.

How did you come at the notion of thoughtful disagreement?

There was a shared personality trait in the people who helped establish the culture of Stripe where they enjoyed disagreement and trying to find the boundaries of an argument and the places it’s not the case, and what the exceptions might be. Trying to establish the topology of that space and trying to construct the map of where different intuitions and heuristics apply and where they might not. They don’t see disagreement as an attack on them, they enjoy finding the limitations of arguments and beliefs.

What would you say is the biggest difference between the Patrick making decisions today and the Patrick making decisions five years ago in terms of how you actually make those decisions?

1. I place more value on decision speed. If you can make twice as many decisions at half the precision, that’s often better. The improving of decision-making with additional time often flattens out. I think most people should be operating earlier in that curve. Make more decisions with less confidence but in significantly less time. In most cases you can course correct. Fast decisions are an asset and a capability in their own right.
2. Don’t treat all decisions uniformly. Consider the degree of reversibility and magnitude. Those with low reversibility and high magnitude you can spend a lot of time on, in the other 3 quadrants you can be much more flexible and fluid.
3. Try to make fewer decisions — why is the CEO making the decision? Sometimes if I have to make the decisions it means something else in the organisation is broken. Pushing others who are the domain experts to make more decisions.
4. When I realise that I would make the decision differently to how someone else is making it, working out what is the difference in our models such that you want to make decision A and they want to make decision B. Currently at Stripe we’re getting different parts of the organisation to write down what they’re optimising for: what their mission is, what the long term key metrics are for their part of the organisation, who their customers are both internally and externally. Once there’s that agreement on those longer term things, then maybe a difference on any particular decision might just be that we differ on what the most instrumentally effective way to achieve this outcome is, but we know we’re unified on what the desired end state is. That way you can quickly test who is right/wrong. The more troubling ones are where there’s more latent disagreement on what you’re actually optimising for but that’s never explicitly surfaced or uncovered. Now in decision-making I place more importance on making sure we have the right foundational agreement so that the kinds of disagreement that tend to arise are of the more superficial sort and where agreement is actually less important.

What do you do at Stripe to make sure you and your people are learning from the decisions they’ve made, both positive and those you would have made differently?

In investing decision-making is very binary, in organisations it’s more about designing feedback mechanisms.

The things to optimise are the incentive structures, the mindsets, the definitions of the goals, and the feedback mechanisms from outcomes to the inputs and the work and the operations themselves. It’s less about the binary decisions.

Organisations are more like cells evolving.

Which company cultures do you admire the most?

Cultures that are strong — where people can expound on the merits for half an hour, or disagree with it. Whether that’s the New Yorker or the military.

Cultures of perfection — The Economist and Apple have extraordinarily high standards for themselves.

Cultures that have longevity and sustained success — The Economist, Sequoia, The New Yorker, Koch Industries. There’s no one thing like the iPhone that enabled their rise, it’s clearly something deeper and more institutional, and has been sustained for so long.

What have you learned from studying Koch Industries?

It’s striking how Warren and Charlie at Berkshire Hathaway and how the folks at Koch Industries are into a kind of epistemology and structuring of doubt, and accounting for biases, and mechanisms for clarity of thinking.

It’s half investing and applied epistemology and half philosophy.

What’s the smallest habit that you have that makes the biggest difference?

I reach out to people whose work I admire and tell them that, and often that leads to a dialogue, and sometimes I’ve got to know them very well.

What would you say is the most common mistake you see people make over and over again that you wish you could correct?

Not having the right peer group or not striving to be more like the ‘right’ people. Whoever the right peer group is will be hugely influential in where you end up. One you’ve found those people you should really invest in it and let them shape you.

Talk to me about how you foresee payments changing from both the customer perspective and the merchant perspective?

There is still so much friction in payments. Micro transactions still aren’t possible because the infrastructure isn’t right and so the businesses that would take micro transactions don’t exist. Or making big payments online. Or the geographic balkanisation that comes from difficulty in paying across borders — 99% of Brazilian credit cards do not work outside of Brazil. These limitations have a big impact on how we interact.

Making it easy for any 2 parties in arbitrarily chosen countries to transact will have enormous consequence for the world.

Also on a deeper level — what determines how many firms there are in the world? What determines the character of those firms? What determines who starts them? These things are moveable — 60% of companies started using Atlas say they wouldn’t exist without Atlas. If Atlas gets big that could have a huge economic significance. Teaching management practices to business owners has been proven to have a significant influence on revenue.

Stripe wants to perturb the overall system to impact these questions and measures, mediated by payments as this foundational level because it’s something every business necessarily has, and because it gives us a good understanding of the dynamics within the business. Even though it’s not about just the payment, we think that the impact of just solving the payment will be enormous.