Tobi Lütke：构建 Shopify 的21年

One of the things I admire most about you is that you've been running your company for 21 years, something like that.

Yeah.

I love people who do things for a long time—people who chase excellence and try to be great. My entire life is founders, right? During the day, I read biographies of history's greatest founders. I've read around 410 of them, and at night, I hang out with founders. I don't think I have a single friend who's not an entrepreneur.

Love it.

I'm always asking the founders I most admire who the founders they most admire are, and your goddamn name comes up over and over and over again. What they talk about is that you have very unique ideas on company building and management, and they use the word “singular” a lot.

Yeah.

I want to lay out how you think about building companies, building products, building technology, and then spreading the gospel of entrepreneurship. I want to start with one thing that you said: “Companies are technologies themselves.” What do you mean by that?

In a lot of ways, I mean, take the social angle. A lot of companies are social technology in the sense that they allow us to go all-in. The only time you're really allowed to spend 14 hours a day singularly pursuing a thing is— I mean, we want kids to spend this time in school, and that's okay, and then in university, you dedicate yourself. Other than that, you can't just not have a job and be really, really into things. It's socially not acceptable.

So company building turns out to be the perfect excuse. Once you call it a company, it's not tinkering around with your ideas anymore, and you get to explore things. What a company fundamentally allows you to do is run the counterfactual to the world you see around you, right? You get to try to build a thing that you think ought to be there, and then you means-test it against the market. If the market agrees with you that this thing needs to exist, it moves energy in the form of money back to you, so you can do more of the thing that you were pursuing all along.

And not only that, it's self-financing. I started a company that I didn't think was going to have a very, very big market, and along the way, the market pulled Shopify out of the project I started, essentially. That's an incredible intelligence to tap into.

So you put all these things together and say, “Well, look, the concept of companies is not that old.” We're on a 500-year run, having evolved out of things called companies, which were actually more like quasi-governments, like the East India Company. The modern company is not that old.

I find that if you take that mindset—that companies are just a path-dependent solution to social and somewhat legal problems—they allow thousands of people to join your project, and it's called a job, and therefore everyone accepts this. Obviously, you can make money, so it's a good deal. Then you can figure out if this counterfactual of yours might be correct or needs updating along the way.

At the end of the day, if you're lucky, you work with other people who are really all-in and inspiring, and who take it further than you ever thought. I marvel at the institution of a company, because if it didn't exist for some reason and someone proposed the whole idea now, it would sound insane. From first principles, none of this really makes sense.

Well, I heard you say something that was fascinating. It really piqued my interest when you said, “We do not know how to build companies yet. All companies are terrible, including mine,” which you said.

Yes.

And, you know, that's—

Mm-hmm.

—hilarious when you're running a $200-billion-plus company and you think that in the next 20 years, we're going to look back at what we were doing at this point and be embarrassed by what we were doing. Why do you think that?

Because it's like everything else. I'm sure even you listen to your first podcast and you're like, “Oh, my God!”

Oh, please don't.

Please don't. But honestly, this should be seen as the most joyous of moments. You should actually do it if you haven't done it.

No, I do it all the time. Yeah.

Okay, good. The difference between that and what you would do today is your progress, right?

Yeah.

As a computer programmer, I experience this all the time. I look at old code and it's like, “What the hell? This is terrible. Why would I ever do this? This is way too abstract and complex.” But I made it better because I now have the skills to do so.

One of the saddest days of my life was when I opened old code and was really impressed with how good it was.

That was one of the saddest days?

That's really, really the saddest day of my life, because I'm like, “Holy shit, the implication of this just hit me like a train,” right?

That you weren't progressing—

—and you weren't learning.

Exactly.

It figures, because I wasn't programming all day. I was trying to build a company instead, which honestly is very, very similar.

This is one of the things Daniel Ek told me about you.

Yeah.

Here's what I'm observing. Like you, I read a lot of books. I try to figure out—if you don't read books, you live a lifetime. If you read books, you live a thousand, right?

I'm going to interrupt you real quick, because I took notes on this podcast six years ago. This is the first time I came across you, and you said, “Books are the closest thing you'll ever come to finding cheat codes for real life. You can access the entire learnings of someone else's career in a few hours.”

It tracks, doesn't it?

Yeah. I mean, I dedicate my life to this, so of course I'm going to agree with that.

Absolutely. I mean, you're predisposed to agree with me, but I really think we need to shout this more from the rooftops, right? The one weird trick seems to be just read books, right? It kind of doesn't matter. Just make a habit of reading books, and ideally change genre every 3 books or so—at least for 1 book—and that alone will give you a range that you can draw on for basically everything you'll ever do.

Look, I read lots of books, especially when I went from being a programmer to being in business and realized I had to learn business really quickly. I got really dismayed with the quality of business books pretty quickly, because frankly, I think business books are largely written by the people who have time, who are not the people who actually build companies.

Yeah.

It depends on whether the person who starts a company or the person writing a book is a salesperson: every problem can be solved with sales. If the person is a marketer, clearly every problem can be solved with marketing. At least I can take that away as a trap not to fall into, right?

I was determined not to be the engineering-type founder who was going to see everything as an engineering problem, because that would be called blindness, and that would cap my capabilities. So I really tried everything, including how to build the team in such a way that I delegated everything, had my different business lines, and arranged things.

This was after the IPO, and I realized I had to be a very “serious” public-company CEO. It almost killed the company, honestly.

How did it almost kill the company?

COVID actually saved it at this point. I had the inklings that something was really, really going poorly. It was a time when we didn't have a lot of competition, which is also really, really hard for businesses. It's not a good thing at all, because you don't have good rivalry, and you can't even tell if something seems really good. There's no one else to keep you honest about it.

You have a distinction between a competitor and a rival.

Yeah. And you think competition is one thing?

Rivalry is really good. I mean, it's the same, but it depends. It's a mindset change. If you compete with another company, there's a lot of, “We also need a version of what they have”—copying, like you do with a Xerox strategy, very quickly. Companies tend to get obsessed.

There are companies where the most active channel in their Slack is a competitive-analysis channel, where people just bring everything everyone else is doing. I think that while that has some merit, the problem is it makes companies very reactionary, right?

In the arts, part of everyone's art studies—or at least in fine arts—is that you copy the great pieces. You make copies of great works. Your next painting is not at the quality of a Van Gogh. You just copy it, right? Mimicry is actually not an excellent way of getting to excellence, and companies fall very much into this.

Whereas if you treat other companies in your space as rivals, it's much easier to have a positive-sum outcome there, because rivalries inspire you to be your best. Agassi could not have been Agassi without Sampras being there, right? He very clearly states this in his book. The Sampras that existed during the Agassi era, that Agassi understood, wasn't a real person. When you read Sampras's biography, he just liked tennis.

Yeah, and Agassi hated it.

Agassi hated it.

And so the rivalry he created for himself created him in a very real way.

In “The Last Dance,” which I loved, Michael Jordan admits that he might have made up a slight at some point, which to me is one of the most profound moments.

Yeah.

Sometimes I play “The Last Dance” in the background while at work.

It's my favorite documentary.

My second favorite one is—

It makes so much sense.

The Defiant Ones.

It's about the partnership between Dr. Dre and Jimmy Iovine—their multidecade partnership.

That's good.

That's the description.

That's not what it's about. It's one of the best documentaries about entrepreneurship.

Interesting. Hip-hop and rap history is incredibly entrepreneurial.

It's fantastic. So, we just recorded with Jimmy Iovine, who's the star of the documentary, 2 days ago—or whenever this was.

I don't even know what day it is.

Amazing.

And that's exactly what he said.

It's like they understood the business—

Yeah.

—of music before anybody else.

Yep.

That makes perfect sense and totally tracks. You can see it from outside. Actually, we see it from inside. It was the first category of music that just killed it on Shopify. Opportunity is converted into products like that, and they're made different. It's really, really cool.

This is the beautiful thing about Shopify. It's a front-row seat to seeing how high-agency the different industries are and how quickly they absorb new ideas. Anyway, it's a different topic, but—

Almost like in a video game—a god-level view of entrepreneurship. How many entrepreneurs are on the Shopify platform?

It's like millions, right?

Yeah. And you're drawing insights for your own work from the entrepreneurs on your platform. One of them we'll talk a lot about today is the importance of differentiation.

Mm-hmm.

But I want to come back to that. I want to go back to this: You were thinking, "Okay, right after the IPO, I'm going to approach everything as an engineering problem?"

Yeah. I'll cosplay a public-company CEO, like a 60-year-old guy in a suit.

With a legal background, right?

Okay. That's sort of what everyone emulates, for whatever reason. And it just worked really poorly for me from a product perspective. The company did fine all the way through.

COVID happened, though, and I'm like, "Okay, so what my company needs here from me is 5 different things." One of them is that all plans are fundamentally invalidated. Everyone who has a plan, please throw it out. We need to go through everything we are doing and rederive it, because everything is based on a very long tree.

You start with axioms, then you make a huge amount of decisions on top, and then you get to a conclusion. That means this is how you spend your day-to-day building this thing because of these things. If any variable along the way is invalidated, what you should be doing is rederiving the entire thing on top, right? You should always prune back the decision tree all the way there and go forward.

People being unable to go out of their houses is one of those things. I don't think we all knew that was a core assumption we had—that people could just freely move around in the world—but we were told that we couldn't do this anymore. It ended up being quite long during COVID, at least in California and Canada and some places.

So we needed to rederive everything, and that was kind of clear to me. That's how you structure a computer program, frankly. I went through the entire roadmap, and what I figured out was, first of all, that was really, really hard to get. People were fighting and resisting talking about everything that was going on.

I found out why: There were just a lot of boondoggles going on. There were random projects by random people that I would have never found out about, all that.

Why did you not know that it was going on in your own company, though?

Because the company was, let's say, 4,000 or 5,000 people. It was quasi-distributed. We started in Ottawa, Canada. We had a Toronto office and a Montreal office, and so on.

It just turned out that, in the Toronto office, there was a fairly big project to build the features to add to Shopify so you could use it for running a supermarket. They decided that the supermarket industry is very large, and we should go and capture 1% of all that, like in the [unintelligible], right?

Frankly, I'm not saying those were wrong decisions. The really surprising thing about building a business—or decision-making in general—is that making the right decision is actually child's play, usually. You can make wrong calls which, because the world changes, end up being the wrong call later, and so on. That's a different thing.

But most of the time, the bad options in front of you just prune themselves away. They don't make sense. They're too expensive or just off-mission, or whatever.

The problem is—and I think school unfortunately, accidentally drills this into people—that "there's one right answer" is what people think. Once they find a right answer that could plausibly accomplish the goals they set out, they go with it.

But there are probably hundreds of great answers, and one of them is far less work. Especially in an engineering project, the best solutions are always modular and component-based. You build one thing, and it will work with everything else you did.

A lot of the bad software that people use are things that are just hairballs inside, where none of the pieces fit together. A huge amount of code has to be written to extend the new thing and make it usable from everything else.

I'm not going to bore you with those kinds of details, but if you look at the project brief, the team killed it. Probably everyone made their bonus. They delivered the thing. It's an island unto itself, and after it's delivered, nothing else in the product works.

So, you were looking at things as "cosplaying like a CEO or a businessperson," as opposed to an engineer?

Yeah. I was trust-falling. I was saying, "Okay, cool, you seem passionate, and you seem to know what you're doing. I trust you to do the right thing. Obviously, we're going to keep talking, and you give me updates on things."

People learn pretty quickly what the projects are that I'm interested in. Talk to me about those and nothing else. People learned how to direct my attention to the things that they wanted to have my attention.

So, how did you fix that?

The moment I went through absolutely every project, I did it myself and took 16-hour days. I canceled probably 60% of the projects, and over the course of a year, I turned over every one of my executives.

This is basically the consequence of this, because I realized that, in many cases, the core skill was—well, I don't want to be unfair, right? It's just that trust was broken in many cases. In a real crisis like COVID—

Again, COVID was hard—the hardest time of my life, for sure. In a crisis, if everyone is a 1 before, some people go to 0 in a crisis, and some people go to 100. I found it's almost unpredictable.

In fact, if I had been betting before COVID about who was going to be contributing the most, I would have been wrong, I think, entirely.

Why do you think that is?

I have been thinking about this for a long time, and I still haven't gotten a good answer. I don't think they would have known. This is the interesting thing.

I think a real crisis tests you in a way that nothing else does, and nothing prepares you for it. At the end of the day, the thing that I have found is that people who can adapt the fastest actually self-identify. Now, I would be able to predict it. Very simple: "Have you started a company before?"

So Shopify is by founders, for founders, causing more founders, right? It's a celebration of entrepreneurship. We believe entrepreneurship is glorious, and people will reach for their heroes and do modern-day heroics, go on a hero's journey with lots of naysayers, overcome incredible odds, and lots and lots and lots of headwinds. That conviction runs really deep.

Sometimes we buy companies, of course. Many of the people we purchased are still at Shopify. People don't stay for a very long time because I think it's such a place for founders and for high-agency people.

I always had a Slack channel with all the founders of the companies that we purchased. I did a founder off-site with them once a year because I find their advice to be so good. They all have been responsible for people's livelihoods, and that changes you.

So I went to the founders' channel and said, "Guys, I need help. It's crazy here."

Had you already started turning over your executives?

Yeah. Once I realized that—

Okay.

It wasn't one moment.

How did you know that instinct—to go to the founders' channel and ask for help?

I think it was a moment of, "Where the fuck do I go? Who do I have most in common with, and who can actually relate to the types of problems here?"

Again, my relation to Shopify is different, right? I started it, right? It's different from everyone else. But the second-best thing is to talk to people who have had this type of relationship with a thing they created.

I asked a lot of the founders to become my executives. In some cases, I took people who were individual contributors on an engineering team and put them in charge of a very large thing. You know what? Every one of those things worked.

It was really remarkable because it fed into an intuition I had anyway. What I found is that, during COVID, the founders of a company were actually not doing well because they were similarly managed. They were irritants.

What does that mean?

They don't settle. They talk about absolutes. If something is shit, they say so. It doesn't matter if everyone has agreed to move on or something. It gnaws on them in a way that's deeper than this other thing.

And it's exactly what you have inside of you.

Exactly. And so I feel it's a very special thing, and companies reject it. Companies cocoon them. Sometimes they give them outskirts, like the skunkworks team; it's daycare for people who otherwise tell you that your shit doesn't smell, right? And shit does smell. So I'm like, “No, you don't get to put them in founder daycare. I'm going to put them right in front of you, or, in fact, on top of you.” And so the difference that made to this business is nuts.

And so this is why I'm saying COVID saved Shopify, because I needed to make a move that was very much, “Okay, I built this company. I got to this point. It's important.” It's actually maybe even more important during COVID than before, because it was, for millions of small businesses, their livelihood during this time and the only way to make a livelihood. And I'm super proud of what we did, because I think the businesses that started on Shopify and came early actually grew and did better.

Some of them grew faster than you. You said somebody beat you guys to—

Yes.

A billion the other day.

Yes.

Which is the craziest thing. When have businesses ever raised their customers that started on the platform?

It's the greatest. I love this thing. And so, again, they're fantastic. They're heroes, they're glorious, they're wonderfully discontent. My customers have the same irritation with my software. And, like I tell you, it's perfect, right? It's exactly what I want.

And so this was one of those moments where I was saying, “You know what, Tobi? Maybe you didn't build Shopify completely by accident, and your intuition might have actually been helpful, and cosplaying someone else is probably not what you need to do.”

At that point, you were running Shopify for how long?

15 years? Yeah, about that.

So you were 15 years—

15 years, yeah.

This is very common in biographies. I think it would surprise people, like, “Oh, why did it take so long to figure it out?” It's like, no, it takes a long time to actually build the skills. And then the important thing about intuition is that it has to be refined.

Yeah.

So maybe in 2005, you shouldn't have been trusting your intuition.

“Trust your gut” is sometimes good advice, but it really depends on your gut.

Yeah.

No one is born with intuition for building businesses. First of all, every decade is different. Some problems are persistent. I laughed when I listened to, I think, the episode you did on Larry Ellison. You were describing his frustration with problems in his sales team, which literally, I had to—it's like, “You know what? I didn't actually have to live that horror.”

I could have actually, if I had read the right—

If you read software—

Or any biographies, or listened to founders. Yeah, there you go.

The incentive problems around commercial teams, it's like—

Yeah, it's just like—

But that's the amazing part. I'm glad you picked that up, because he's like, “I'm fucking embarrassed.”

He's like, “I'm 40-something years old.”

Mm-hmm, yes.

Yeah.

“I've been running this company forever, and I didn't understand that incentives with my sales team mattered,” and they almost destroyed his company.

Yeah. Yeah.

They almost went out of business.

I usually listen to podcasts when I'm sitting in my racing simulator to practice for my next race, and it got to that point, and I had to go, “Okay, I have to stop and just listen to this, because I'm just feeling so much empathy for Larry Ellison of all people.” How funny is that? So, anyway—

Who's that empathetic person? Before you move on, because I love where you're going. I can't help myself, because you know this. It's like, when I hear you saying, “I'm getting rid of the executives. I'm going to build an executive team of founders,” that's not a new idea. Rockefeller was doing that 150 years ago.

It's interesting how the building blocks that Rockefeller largely invented—I mean, we call these things trusts, because that was a term—

Mm-hmm.

—they first used there.

So now, you have the executives. You're going through the most difficult point in your life.

Yeah.

And you realize, “Oh, I'm cosplaying. I'm going to go lean into my intuition.”

Yeah.

This is another reason why I think that decision you made 6 or 7 years ago, however long it was, is why so many founders study you, respect you, and take pieces of your philosophy as well, because you're like, “Goddamn it, I'm just going to do this the way I want to do it.”

I was like, okay, I need to try something different. You have to do things differently, axiomatically. Obviously, axiomatically, if you do the same thing, you get the same results. Differentiation requires new ideas. Instead of spending the next 10 years of my life figuring out how well I can cosplay someone else, what would a company look like if it just taps entirely into my intuitions and my biases?

I'm doing literally the opposite now of what I did before, because this didn't work, so maybe the opposite works. Worst case, I triangulate the midpoint between those things. So, enabled by the very clear, fast, and obvious success of asking people from the founders channel to step up and run the company with me, my real set of intuitions here—

Again, intuition is your entire life's story rolled out for you to make a snap decision in the moment. What does my intuition say about what a company actually should look like? Let's forget about the path dependence. Let's go full tabula rasa and go up from axioms. And I started with really, really, really nothing. I started a project on GitHub, literally.

So, engineering tools—my intuitions again. What is the first principle of a company? What are the inputs? What are our decisions? Put in config files: what titles exist, what levels exist, how—

I asked for all the sales compensation data, all the market data, and put them into the repository. I turned them into machine-readable files because I got them as PDFs, and so on and so on. So I built a thing that, at the end, had all the information to create a model of the company.

Like, literally, this is Python code that takes the configuration files that we agreed on—things like how many people should report to a manager, all these bits—and then computes them. It uses something called a solver, like a SAT solver, specifically. And then, with all these constraints and all these inputs, what should Shopify look like? What departments exist? What level exists? How many people are there in which group and everything? And, of course, the first version was utterly incorrect, and then I realized—

Are you doing this alone?

I started it, and then I got 2 or 3 other people involved to help me with this, and we really got into it. And so this is called Shopify OS, the Shopify Operating System. It is really a large part of Shopify now.

What it did was make it irrefutable that we didn't know all the principles. There were 8,000 people in the company, and there were 5,500 different titles in the company. Some of them are just senior and senior staff, but in some groups, senior staff was lower than director, and in some, above. And when I tried to make something software-addressable and tried to reproduce it, it showed every crazy, unviable choice in the company very, very easily.

So there's a system in engineering—not actually a well-understood part—but I think it should be called desired-state systems. And what desired-state systems are is a thing where you say, “Okay, here's what should be.” You hold that up to what is. Like, you click on a link on a website, and your React library behind it, what it does is say, “Okay, I compute what it should be, I see what it is, and now I figure out what is the minimum number of steps I take to get this to that.” And that's what you see in your web browser.

I need a desired-state system for this company. So we have this model we created, and then we hold it up to the company, and the job of HR is to be this reconciler.

How do you take the minimum steps to get from here to there, or change what we have in this system to approximate what we actually want? This is very technical. Here’s the effect of it: think about how much politics this removes.

When my head of sales comes to me and says, “Hey, I need 50 new salespeople,” I can put this into the system, and it says, “Well, that means you have to either make these changes, or you’re going to lose some of your engineers.” Because the system, based on our agreements, recomputes everything.

So we can always look at the counterfactual. It’s not like, “Yes, the boss agreed while playing golf to hire 50 more salespeople, and now someone has to carve—you have to find a pound of flesh somewhere in the engineering team.” Suddenly, you don’t actually work on innovation anymore because you never made the decision not to hire engineers because your budget doesn’t allow it anymore. But you did make the decision to hire salespeople.

Maybe this is the right decision, but now we can actually look at the consequences of every one of those things. It was hugely successful because it created such legibility and simplicity for us. It also told us that so much was undecided.

For instance, we took the concept of— in engineering, this was already around—individual contributors are a really excellent career track. You can make as much money in good companies as an individual contributor as you can as a VP or manager. The best-paid people in many companies are now individual contributors, especially in machine learning, at least at machine-learning labs like OpenAI.

But no other discipline really had that. You had to become a manager to progress in your career. One of the consequences for us, when I really looked at it, was, “Hey, I need more great engineers.” The best place to recruit them from was my own management team. That seems silly, especially because are they actually great managers? They’re really well respected because they’re good engineers, so maybe, but don’t they want to code?

So we created this sort of mastery system, we call it, where every job has the ability to stay—you can get these level upgrades frequently, and at any set of responsibilities, if you’re insanely good at it, you can make lots of money. That just seems like, of course, you design it like this. Why would you not?

So they can stay individual contributors.

Yeah. Take an example: in engineering, if you’re 100 times better than the other people in this discipline, usually you can make more money than a vice president. That depends on the value you can bring, but it is possible.

I kind of hated the way we did compensation in the industry. Again, these are boring topics.

No, it’s not. They’re not boring to me or to the people listening. Wait, why do you hate the way they were doing compensation in the industry?

What I love about entrepreneurship is the level of personal responsibility people take. I said as much earlier when we talked about the founders who have been responsible for people’s livelihoods. It changes you.

It does, right?

So I don’t work well, myself, in an environment where everything is prescribed. I don’t like joining other people’s games, and so on. All these kinds of things are my biases, and again, we are implementing my biases here.

So in COVID—this was 2022, I think, or 2021—suddenly Shopify went down 80% in stock value. Again, I think that was a problem because I think the stock market is basically a polling market. It’s sort of a betting market on a future value, right? When you guys are wrong, you’re just bad at betting.

I work on the real market value of a company, which you guys are betting on, but I don’t know. There is such a discrepancy between the way I see people talk about stock prices and how I have ever perceived them inside companies.

But what was going on in your head when it dropped by 80%?

Well, in my head, I’m relieved because I’m like, “We have a stock loss.”

Relieved?

Yeah, of course.

What do you mean, “Of course?” That’s not a normal reaction.

I was not about to raise money. I had enough money in the bank. I’m German, so I don’t take on much debt, and so on.

Were the internal metrics strong?

Yeah.

You felt you didn’t deserve that?

Yeah, the unit economics are amazing. However, I don’t want to say this, but I honestly don’t remember the exact numbers. At the high point of our stock price, I think we were trading at beyond 50× revenue and stuff like this. I’m like, “Yeah, that’s not exactly value investing here, right? Guys.”

But you weren’t worried about the fundamentals of the business?

No, but when it is 50 times revenue, I see it as my obligation to do that.

Okay.

My job is to build the best company I can. The betting market says that I can deliver that kind of thing, and I’m like, “Yeah, maybe, but you’ve got to give me a moment here. I need a moment here to do this.”

The reason I bring this up is because I talk to another entrepreneur and I just run it through whatever the hell’s in my head from all these books.

When I hear you talk about that, I think of the time when Amazon stock—Bezos talks about this in his shareholder letters—it went from something like 180 down to 6, or 120 down to 6. He’s like, “Yeah, but they were focused on something else. I saw the actual fundamentals of the business, and I was like, ‘This thing’s going to work.’”

Exactly.

I think Buffett always repeats that thing where, in the short term, the stock market’s a voting machine, and in the long term, it’s a weighing machine. You just want to build a heavy-ass company.

Yeah, exactly. Jeff’s like, “I’m building a very heavy company. Amazon will be a heavy company, so I’m not worried about this short-term drop by 80% or 90%.”

That really tracks. But one of the effects was that people got stock options at that level, and they rightfully said, “Well, how about it?” Even if you say, “Okay, I’m super game here to get us back there,” this is going to take some doing.

They realized the psychological problem with everyone being so underwater and having to spend years getting to basically the zero point, where the stock options are worth even a penny again. Of course, they were like, “Well, but that’s the company. The company gave me those things at the time. I received them. I was passive in this interaction. Therefore, the company has some kind of responsibility to make me whole here.”

Yeah.

I think that’s actually—I mean, I don’t necessarily see it this way because the risk was there and these kinds of things, and again, it’s other people voting that’s causing this price. I didn’t set it. However, I do receive the point: they had no agency in the process.

So we rebuilt our compensation system to be completely the opposite of what everyone else does. We give people, “Here’s your annual salary. Here’s an internal system. You go in, you look at it. This is a number. Now you get sliders. You can say, ‘How much do you want in stock? How much do you want in RSUs? How much of this do you want in Shop Cash, actually, and how much do you want in cash?’”

You can change it every quarter. You decide how much money you want. You can even use a tool that’s there to lock in the value of the stock you receive for 3 years.

Sometimes orthodoxy can come back on the table if you get there from good principles. You can actually join Shopify and get exactly the same stock-option deal that people get at other companies by using the tool we give you, but you have full agency and you make this choice.

The consequence of this compensation system is kind of beautiful because, first of all, it’s super predictable. Second, if you hold stock and it appreciates, you make more money. If the stock goes down, you now get more stock options in your next quarter. So it’s rebalanced against the actual value every quarter.

It works really, really well. It’s very popular. It was kind of hard to do because doing this worldwide with people in many different countries was a legal nightmare. There are all sorts of rules around changing salaries. We figured it all out, so we have a blueprint. If other people want to do it, it works really well for us.

We’re proud of it because, again, it’s a point of differentiation, and we believe it works much better. I want people at the company to also feel like this is a company that never sleepwalks into anything. We are deliberate about things.

When you started this—“I’m going to do it my way. I’m working with the founder executive team”—you were in the process of that, and the stock was dropping?

No, it was already there. That was an early 2021 project, I think, and then I think the stock market—when did that change? Late 2021? These years are totally running together for me because there were no weekends. There was no everything else.

I can’t tell you who said this. I talk to a lot of older founders. People think I’m obsessed with old people. It’s not that. It’s like, this guy’s got 50 years of experience as an entrepreneur. Turns out they know some shit. They’ve seen some shit.

“Surprise.”

You’re going to remember the beginning of your company, you’re going to remember the end, and so much in the middle is going to disappear. He said exactly what you said.

It just blurs together.

Yeah. So he's saying you should document it more.

I totally agree. Lots of years are totally running together. I have reasonable record-keeping on this, and especially with all this AI-led agentic reprocessing of notes and press, you can actually reconstruct these histories and timelines. I'm fascinated with this right now.

So my understanding, though, is you're now looking at running a company as an engineering project.

That's right. So I'm like, “Let's engineer a company.” “Be a company engineer,” as I told my team.

Does that apply to every single department in the company? Do you engineer marketing? Do you engineer everything?

Yeah.

Everything?

Everything. I require my executives to at least tell me—but actually do it: go to a podcast, an international conference, and talk about how Shopify does the thing the conference is about differently and why that's better every year. If they don't have a good answer for how they are doing this right now or how they would, that's what I'm working on with them: crystallizing that message.

There are a couple of reasons why I do it this particular way that we can get into.

No, let's get into them now. What's the psychological reason?

I study the brain quite a bit, and the main conclusion you get to about the brain is really that the brain is a retrospective narrative-alignment mechanism. It's actually terrible at record-keeping, but it mostly attempts to take the most salient version of our self-identity and reconcile the history with it. It sort of rewards you for actions that are directly congruent with this identity, and it dissuades you from dissonance with your own identity.

But you can actually change your identity quite a bit, and that makes this whole process work for you significantly.

What do you mean by change your identity? Like, the way you view yourself?

Literally, affirmations work, right? Affirmations work, which is the dumbest trick that works.

I'm so glad you're saying this, because people from the outside are like, “This guy's very logical, engineering...”

But you're all about intuition. Now you're talking about affirmations. You're speaking to the choir, just so you know.

I'm a toolmaker. I always project as very idealistic, but I'm actually not that idealistic. I'm a craftsperson, a toolmaker person. My entire life was making tools. Literally, Shopify is a tool. I make tools very often for toolmakers. My hobby is making tools for engineers or people who make tools, right?

My life's a fractal of toolmaking. It's all about toolmaking aesthetics. It's all about the craft behind it, but not because of craft for its own sake, although I appreciate that too. Even that's not for idealistic reasons, because appreciating the craft behind toolmaking happens to be pragmatically an excellent way to get really good at toolmaking. Everything is about outcomes.

What's the affirmation part, though?

The affirmation is that if you tell yourself or write down 100 times something about yourself, that writes it into the prefrontal cortex at such a deep level that your brain will start reconciling you to that. It just works. I've used that very often. I was terrified of public speaking until I sat down for a week and, every day, spent 10 minutes writing that I like public speaking. “I love public speaking.”

You wrote that down?

Yes.

And now you love it?

Yeah, a week later. This works when you know what you're doing. It's not like a placebo. It's actually like you just actively change your prefrontal cortex in this moment. So I use this often. I sort of write to myself—message in a bottle—and these kinds of things. That's all a different topic.

No, no, no. What's the topic there? Hold on. You write messages to yourself?

Yeah.

Affirmations or just generally?

No, just generally. When I feel like there's something I've figured out that I really like, I know I have to write a scheduled message to myself to remember it, because I need spaced repetition on the idea.

What's an example of that?

For instance, we have a company keynote kind of thing—a company get-together summit once a year, and it runs a full week. Usually, I do a keynote there, and then I write a message in a bottle to myself about what was good about the process, what was bad about the process, what to do differently next time, and all the topics left on the cutting board, just to get a head start. That's a random example of that.

How did you learn that you could change your identity, that affirmations work, and that writing these things to yourself works?

I think trial and error, but I think the hints are everywhere. Affirmations are one of those things that—I mean, there's a reason why society creates rituals, right? Because rituals end up being largely affirmations on repeat.

What do you think of visualization?

Manifestation and these kinds of things probably work too. You know what? The funny thing is, I think everything works if you just do it. The question is, what's effective and quick?

It sounded to me, before I started reading all these biographies, like, “This is some woo-woo nonsense.”

Yeah.

Yeah. And then you see very smart, analytical—

Yeah.

—logical, in many cases, inventors and engineers, and it appears over and over again. People didn't know each other. They worked in different industries, at different times, lived in different parts of the planet—

Yeah.

—and they all talk about doing this.

And it's parallel construction. They all come to the same conclusions in different words.

Yeah.

So, yeah, personally, in the self department, I'm extremely taken with Stoic philosophy: control what is under your control, know what is, and then do the best possible job. The job is the thing. The whole project in life is just to try to become very, very good at something and, ideally, cultivate some significant skill, create potentially useful services and products, and then share them with as many people, right?

You do this because that's what you're on planet Earth for, right? Do things that matter in the world, that are under your control, but also sign yourself up for the most interesting places, as long as you remember what your job is within them.

I love the current times, and I love the technology industry. I have a role inside of it, which is taking care of millions of entrepreneurs and small businesses that are trying to use the world of technology and the world of retail through their own work, then self-actualize. That is a worthy goal and a worthy job.

So AI comes out, and people are like, “Well, should I start an AI lab or whatever? Don't you want to work at OpenAI?” Everything would be interesting, but you know what's really cool? Pursuing the same problem in a changing landscape. Now I get to reinvent everything I've already done. How fun is that? I really care about these problems.

Anyway, back to the affirmations and the brain's self-gymnastics. Going to my team and just saying, “Look, just tell me what you will talk about on stage,” is that. Afterwards, you need to internalize this as something that you have committed to me, as something they care about. They care about reconciling, and so they have to do it, because otherwise they said something untrue to me, right?

They don't see themselves as someone who says untrue things, and therefore, you can actually make the reconciliation work for your own benefit in this kind of way. It also just means it emphasizes difference, right? Difference is so important. Orthodoxy just really needs to be off the table from the beginning of a project or from the beginning of decision-making.

Have you ever studied James Dyson?

No, no.

If I can recommend only 1 book ever, it's his first autobiography. He has a crazy line about this. This is also very common with great entrepreneurs, and you hit on it in your earlier talks: if I'm trying to chase excellence, trying to be the best in the world or make the best tools possible, therefore, that demands difference.

Yeah.

That path, that mission I'm on—I can't do it if it's the same. Literally, I can't be the best if it's the same.

Yeah, right, exactly.

Just think about this.

Yeah.

Edwin Land, who was Steve Jobs's hero and the founder of Polaroid, had a personal motto. He said, “My personal motto is very personal, and it may not work for you,” but it was, “Don't do anything that somebody else can do.”

He just would not make a “me-too” product. But the reason I bring up Dyson is because now he owns one of the most valuable privately held companies in the world, right?

Yeah.

He's writing that book when the business was fine, but he had 1 product in 1 or 2 markets. It was doing around $300 million a year in revenue—not a bad business, but a very small business compared to the size that it compounded to over the next 4 decades, or 3 decades. He says in that book, even at that time, “Differentiation and retention of total control.” And he goes, “You make it different, even if it's worse.”

Yeah.

That's one of the craziest sentences I've ever read.

I totally agree with that, though.

It has to be different, even if it's worse.

I completely agree. I see it all the time because, again, it gets back into the copying-of-a-painting problem.

You can make a 7-out-of-10 solution to everything by copying the 7-out-of-10 solutions that are already out there in the market. But if you make your own version of it from tabula rasa axioms, you have mastery over it. Even if it's a 6 out of 10, you can iterate on it. You can take this past the 7 afterward.

Very often, especially in technology, the world belongs to the fast: the people who iterate, adjust, understand what's costly and what's unnecessary, and prune away the rest. I find that, probably pixel for pixel, the most inspiring picture that I think exists, that I've ever come across, is the SpaceX Raptor rocket evolution. Do you know the photo I'm talking about?

Yes, I know exactly. I use that as an image for the Dyson episode I made.

That's perfect. It's perfect, right? You look at this and it's like, "Yeah, that's today's Picasso, right?" This is it. We're not at a point in time where the great works and masterpieces are being done by 1 person. They're done by teams. Very few teams can move forward by subtraction. That does not exist in industry, usually—very rare.

So it's beautiful in that way. You have to do things differently, make them your own, and have mastery over the first version if you want to take it further. We talked earlier about the podcast I do inside a company called Context, and it's about that. How did we make the decision? Everything around the decision is more interesting than the decision.

Is this podcast constantly updated?

Yes.

So it's internal—

Mm-hmm.

Only for Shopify. Goddamn it, only for Shopify.

Especially when we talk about podcasts, that makes sense. It's like...

Well, they have... It's funny, I looked at my end-of-year wrap on Spotify—

Yeah.

—and one of the podcasts I listened to the most, which was shocking to me, was the internal company history.

Oh, yeah.

It's called Spotify: A Product Story.

Uh-huh.

Because everybody else... I think one of the reasons that—

Yeah.

—we have a mutual friend, obviously, in Lulu. She's on your board, and she talks about the importance of telling your own story.

Yeah.

We went to dinner one night, and I was like, "Lulu, what I love about you is that you're repackaging old ideas." Napoleon told his own story. Julius Caesar told his own story. All these people did it. It's a very valuable idea.

And so, everybody who's trying to tell the company history outside of the company—

Yeah.

—you say, "No, you got all this other stuff wrong. We're going to make it ourselves." I think it's a 10-part podcast. Originally, I think they built it for themselves, and then they published it.

I didn't know about this. I love this.

You should listen to it—

Yeah.

—because it's narrated by Gustav, who was Head of Product—

Yeah.

—who's now co-CEO.

Yeah. And they take it in chronological order, so Sean Parker played a big role at the very beginning, right?

Yeah.

He's obviously not there now, but he's on episode 1 or 2—

Mm-hmm.

—and then they interview the people who made the decisions, and that's the whole point.

At the time, yes.

Executives and founders.

It's totally right.

Not only what they did, but why they did it. And the why is so much more interesting. One of the failure modes of "Tobi said" is real inside Shopify, right? People just throw these things around: "Well, Tobi said we're doing it this way." I just try to get people to always ask, "When?"

When I said it. Exactly. You change your mind all the time?

Yeah, and this is not even... I learn about this because something is really odd—really odd decisions. There were easier ways available. How did we get there? "Well, everyone said that you wanted it this way." "Under which context? When did I say this? Clearly, we've never talked about this." So it's this very common thing.

What's a typical episode? Is it just you explaining the thinking behind what's going on currently in the company?

It's like taking people back to major decisions.

Are these short episodes?

No. Initially, we went through phases. For the first maybe 50 episodes, we said, "We have to mix it down to 23 minutes," because we calculated that that was the average commute in Shopify. So we checked where everyone lived compared to the offices, ran the data, and it was 23 minutes. We made it 23 minutes because that seemed like a good reason. We needed a reason.

And you like designing your constraints. You talked about this a lot.

Exactly. I also just don't like picking 20 minutes because it sounds good. I need a better reason. I might say, "I want something to be... 20 minutes sounds right," but let's find an actual good reason why it's 20 minutes, and then we found a 23-minute reason. Sometimes you're allowed to just parallel-construct the thing you want to do anyway.

At some point, we abandoned it because the nice thing with podcasts is people can skip and move around. It doesn't matter. You can just keep going for hours and hours.

Do you have any that are long?

I think there's one that we had to cut over a couple of episodes that ran like... If you ran it all together, it's like 5 hours.

What was that one about?

That was the philosophy of engineering with a bunch of our distinguished engineers. You would not believe how immature engineering is as a discipline. There are so many good ideas and so many bad ideas, and it's so hard for people to sort through them in a way that's much harder than in any other industry, because software has a weird property.

The best thing, and why the business model of software is so insane, is because copying software at zero marginal cost is free, right? We build Shopify, the first person signs up, and we make money. The second person signs up and uses exactly the same thing, and the second instance of it is not a new piece of software; it's the same piece of software. Hence, the margins of software are incredible.

The same thing causes a lot of problems for the creation of software, because you just don't know how much dead weight is in the system. If you compare this with electrical engineering, there's a very clear cost of manufacturing, right? The bill of materials is the thing that dominates the success of a product, because it dominates the cost, which then dominates the product-market fit.

Manufacturability is also a huge component in electrical engineering. Is the board laid out in such a way that it's prone to defects? Did you choose all the right components? Does one chip get really hot next to a soldering thing that... I remember the original Xbox desoldering itself if you just played certain games for too long. There are real constraints like this, so excellence is obvious in a way. It's in the numbers, and everyone's trained on costing waste into the system.

Software engineering just throws away so much of the capability of these incredibly powerful machines, partly because it doesn't cost anything. However, at a certain scale, it starts costing, but at that point, it's being fixed by operations teams getting more servers, rather than engineering teams going back to fix those things.

The consequences of bad decisions in software engineering are an externality that doesn't accrue back to the engineers who caused it in almost all instances. So it's a little bit like a factory polluting the environment. It just doesn't cost back to them. Therefore, it's a very tricky industry to move forward. It quite often moves backward significantly.

Computers got much faster, and they threw a huge amount of capability away, making these very heavy client-side JavaScript applications that took over everything. They were godawfully slow, very error-prone, and took forever to load, especially in e-commerce, where people visit sites rarely.

Some of our customers went out of business because they talked themselves into their own stack. They hired a CTO who brought in their own team. They used a set of technologies that they wanted to use because they were currently the "in" thing, with the most interesting conferences in the most interesting places.

Then they deployed it, and it sort of worked in the demos. But when you're going to a website for the first time, you have to load it all, and you don't have an up-to-date MacBook with perfect hardware and lots of memory, like everyone who works on engineering teams. You might have an Android device, and you might be on a train right now. You're sitting there for a minute and a half waiting for this online store to load. Guess what the conversion rate of that online store is?

Zero.

Yeah. I'm quite annoyed by this, although that was a real problem in 2023, especially, when I think a bunch of us older engineers had to take the kids aside and have a good conversation.

And that was the context—

Yeah.

—of the episode of Context.

At least that's what led to this particular thing. I don't know if we actually succeeded there, but I think we didn't need to, because I think the world just changed. The moment of the zero-interest-rate phenomenon, when they're gone, people actually wrote better code all of a sudden. It's amazing how consequential the absence of 0% interest rates actually ended up being. You could feel it in the engineering team being less prone to accepting random waste into their stacks.

It's unbelievable. Oh, that's interesting. I don't know if I've ever heard anybody else make that point. I think this idea of hearing directly from the founder in an audio way is underutilized.

I did this episode on the founder of Kinko's. He was dyslexic, okay?

Yeah. So am I.

Yeah, I heard, which surprised me. Actually, a lot of great founders are dyslexic.

It's very common.

Yeah, it's very common, so it shouldn't surprise me. He's like, "I can't read your goddamn email." He also—I think he said he had ADHD—so he couldn't sit down at a desk. His job was, "What do I like to do best?" And he's like, "I just go around..."

At the time, all the Kinko's stores were owned by individual people, so it wasn't one cohesive corporate structure. Therefore, there was a lot of experimentation going on. He would travel the country and spend all his days in the stores, finding good ideas because all these people were experimenting through trial and error.

Yeah.

And so what he would do—this was back in the day—was call into a corporate voicemail. At the end of the day, he'd give a 3-minute update about all the good ideas he saw.

Mm-hmm.

Then you'd come in as a Kinko's employee the next morning, and you could play Paul's voicemail.

That's cool.

I love it. And then this is an interesting thing that you might agree with, though. After doing this for a while, he's like, "You're resurfacing all these great ideas, right? This guy in Memphis found the best way to do X."

Yeah.

"Why don't you mandate that everybody in the company—"

Yeah.

"—does X this guy's way?" He goes, "Because if I do that, that's the best way it will ever be."

Mm-hmm. Which kind of relates to what you just said.

Mm-hmm. He's like, "Listen, there are probably 100 different good ideas or right ideas, and if I stop here and don't keep tinkering"—going back to that word that you like—"then that's the best it will ever be."

If you need something and you tell people what to do, you might get it, or you probably will get it. If you don't tell them what to do, you might get something you cannot possibly imagine if you have good people. This is actually a really important thing. It's much more important to create an environment in which people can be their most excellent selves than it is to prescribe, "Here are the precise moves to make."

So, how do you create an environment in which people can be the best version, the excellent version, of themselves? I heard you say—is this related to your idea where you said something that made me chuckle? You said, "I don't do corporate babyproofing."

Yeah. Exactly. Absolutely, it's the same idea. The babyproofing thing is real. What are proper internal policies? What are processes? They are, "Do this thing differently from the way your intuition tells you to." Okay, why do you want people to behave differently from their intuition? Let's look at the higher-order bits there.

What other opportunities do you have? Well, you could change the environment in such a way that doing the right thing is intuitive. That's pretty good, because now you don't have to post the policy. You could also hire people whose intuition is better honed.

Very often, policies and processes are a sort of downside protection. You're trying very hard to avoid people who do things wrong causing so much damage. That's the downside protection, but you're also really limiting what the best people can do.

I think of creating an environment almost like a box for people: "Here's the space, and inside this space is the problem I need you to solve."

Right now, my world, in the commerce world, is attempting to figure out what agentic commerce is, how it works, how all the companies are working together, how the research labs work with everyone else, how to expose product catalogs from the entrepreneurs, how to make sure that transactions can be made, and what the large language models need in terms of information to do a fantastic job when people ask them to be their personal shoppers.

That's a box. Inside of it, somewhere in the dark, is the most beautiful solution. What tools do we have to explore this box? It's dark. I don't know. I can take a stab at it. I can paint a picture of what I think is good, but that's just sending people into the room with a couple of tools.

Afterward, they need to make their own decisions, and I want them to do this. I'm not there. I don't want to control every decision. The company makes millions of decisions every day in a building like this.

What I can decide is to help them really want to be on this journey, help them fall in love with the problem, and see the potential. I can make sure it's the right team, everyone has the right skills, and they know what they have decision power over. I can also make sure they know how, at some point—because people are only willing to take so much risk—to transfer that risk to me, ideally, or to a company.

In our projects, this takes the form of working in phases. A prototype phase is what I just described, where you explore the problem. Then, when you've learned what you need to learn and think you have something that you can build, you make a proposal and transition out of the prototyping phase. That's a meeting. That's what we call a phase transition in our project system.

There's software that manages all this kind of thing. They can use an AI to preview what the transition will be like because it's trained on all the other reviews I've ever done. They can mock-do it and figure out what I will probably say. That saves me a lot of time.

Then you have an actual meeting, and they tell me what they're planning to do and what they've learned. That helps me a lot. Then I say, "Okay to transition." Some engineering leads and design leads give the first okay. I give the second okay.

Then they're in a build phase, and a new phase starts. That's a new box. They have agency to make decisions in there. At this point, I assume the risk. It's a trade-off of accountability for autonomy.

That's how we found we can almost operationalize this system. What I'm finding is that, in all the biographies you read, the moment of insight—often the product that started it all, or the artworks that define the careers—did not really come out of an office-building environment, where someone had OKRs for a precise metric to hit.

Leonardo da Vinci probably did not have the Vitruvian Man as an OKR given to him. You kind of have to create a box for people, sometimes a physical space as well. He was in an attic; that just ended up being where he did most of his creative work, too, I believe.

You try to figure out what sort of environment allows excellence and creativity to come together, and then collapse that into saying, "Okay, now the company needs what it needs, because we need to make resource-allocation decisions." You're putting more people onto a project. At this point, there are a couple of things that need to be decided, and then autonomy is regained until the next phase, which is releasing, at which point we do that review.

Sometimes we say, "Okay, we're not ready to transition because there are some things we need to do differently." But that's the way Shopify orchestrates now. These are all things that exist today based on the days during COVID when I reviewed every project.

I thought, "If I were a competent CEO, what would I have put in place so I could do what I needed to do at the beginning of COVID very quickly? Given that I wasn't, and now I aspire to be, what needs to be created?"

Whatever the answer to that was, we worked hard on it, and I think we discovered some extraordinarily powerful systems and mechanisms on this journey.

That was the prompt? What would a competent CEO do in this situation?

Yeah.

I love to take the view that I'm a corporate raider, Shopify went bankrupt, and I bought it at a fire sale. I'm marching in on day 1, and previous management was crazy, and we need to turn this place around. That's my favorite. I love playing this game every year.

I figure out what I would do differently, and it's like, "Let's go!" I have my to-do list. I ask a lot of teams to do this. When we have systems that just don't work, one of the most powerful things you can do—maybe through affirmations, maybe through experience—is simply not to be too prideful about the prior work.

You're allowed to have pride in it, but you want to wake up better every day, right? You want tomorrow's version to be better than today's, and for that, you have to downgrade what's there. The sunk-cost fallacy is so powerful.

But how is that an affirmation? How does that represent what you're doing? How do you do it when you're actually talking to yourself or writing this down?

Yeah.

Or is this your inner monologue?

I literally write hit pieces on the past. It's just all the things that are wrong with the thing I'm somehow really proud of.

A hit piece?

Yeah.

You are so much like some of my favorite founders. Dyson's like this. My friend Karim, who is one of my closest friends—

Yeah.

—is the co-founder and CTO of Ramp. It's like nothing's ever good. The way I describe him is just—

Yeah.

No resting on laurels, no sleeping on wins.

Do something great, and then do it again.

Yeah.

Steve Jobs said this. He's just like—

Yeah.

"What's the proper response when you make something wonderful?"

"Do it again." Yeah, let's go.

No, a lot of these great people don't really have rearview mirrors. They'll obviously learn from what they did, but they're not like, "Wow, look at my trophy room."

Mm-hmm.

They don't think that way.

Mm-hmm.

Look, nostalgia was put on death certificates in the 1800s, right? People died of nostalgia in their minds. They really died of something else, but we used to know that nostalgia was not a good thing. It was one of the bad indulgences. Now it's barely a vice, so I think nostalgia needs a little bit more scrutiny.

Overcoming the sunk-cost fallacy is one of those things where any path you can take to get there is going to be enormously beneficial.

Well, I think playing the game of corporate raider is a great way to do that.

Exactly. So then you develop tools for that mindset.

You kind of separate yourself—

Yeah.

—because that was your decision.

Yeah.

You might be like, "Oh, well, I feel a little sensitive about that."

Yeah.

It's like, "No, no, that was somebody else's."

Yeah, yeah.

This is a new company now. So one of the things that is unpopular when I do it inside of a company is that I really shit-talk the past, especially if I did it.

You shit-talk the past publicly?

Yeah, I trash the past. I make fun of all the ways that this is hilariously wrong in funny ways. And I need to remind myself, especially when I work in an area I didn't work on, that I have to provide context to it. I need these corporate-raider tools. I'm collecting these now.

It's almost like dissing yourself.

I need to help people not think that what I'm doing is passing judgment on their prior work, because I'm not.

Well, you are passing judgment on the prior work. You're not passing judgment on them as a person. This is what Rick Rubin talks about in his book all the time.

But it's not theirs anymore once it's out there, right? This is the thing. They need to disconnect from it. The work is collected. The moment the pull request is accepted, you've converted your craft into something for the company. It's in the company's commons.

A company is a collaborative project where everyone contributes, and there's a process by which it's contributed. At that point, the company accepts it as being up to quality and says, "This is now part of the company's embodiment, essentially—the product." You've contributed it. You deserve the credit for this, and so on. But the company could have had someone improve it at any point in time, right? It's not yours.

We don't allow people to use the word "ownership" over parts of the codebase or anything like that. We only allow "stewardship" as a word, just as a proxy term, because sometimes you need to be able to talk about it this way. Shopify thinks of itself more as an open-source project. You're contributing to the commons. People talk about Wikipedia, not about the particular editor of a Wikipedia page. You know what I'm saying?

I have a selfish question before you move on. You're looking at all past work. There's this guy named David Ogilvy, who is one of the greatest advertising-agency founders.

Mm-hmm.

And he says that we have a divine discontent with our work.

I love that quote.

It's a perfect antidote to smugness, right? Because he's like, "You don't want to rest on your laurels."

Yeah.

I talked to James Dyson about this, too, because he has a very interesting organizing principle. So we're going to take this cup right here. This is a product. He organized his entire life around this. He's been an inventor and engineer for 55 years now, and he's just like, "I see this product, and I ask myself, 'How can this product be better?'"

Then I make it better, and I put it down. Then, a short time later, I pick it back up. "How can this product be better?" I make it better. I put it back down. And he goes, "I see the bad in everything."

Yes.

But the interesting part is that he's intellectually and emotionally mature enough. He goes, "You have to understand, it doesn't mean I'm a miserable person."

Yeah.

"I just look at it like, this is a problem. That problem is an opportunity—"

Yeah.

—and I want to work on that. What is your inner monologue when you're critiquing your past work? Are you harsh on yourself?

I get excited.

Oh, you get excited? Oh, shit!

I get excited by finding problems. I love figuring out if we're bad at stuff. At this point, I'm such a long-term veteran of tech-company building. We have a pretty good thing going on. By all intents and purposes, this thing's working, right?

So when I figure out that we're actually bad at something, I'm like, "Holy shit! I have an obvious blueprint now for how to become a better company," which is actually my job, my craft, right? These are all opportunities.

Things go into the commons. They are hopefully being judged by everyone. I get grumpy when things that are shitty are not identified as such and improved. Then I make people go do it. Sometimes I do it by trying to create energy by just roasting the thing's current state. I want people to feel no deference to it.

I don't want people to respect the piece of code that exists, or the system that I want them to improve. I want people to find their energy source, because what I'm asking them is, "Go into the particular box that contains this problem and do the best possible job—better than anything I could possibly ask you right now—in solving it."

I expect people to be highly agentic, fall in love with problems, and find better solutions than what the industry usually produces. That's the entire point. It's unpopular when people aren't in the right headspace for it, and it's incredibly effective if they are.

The difference very often is just a couple of words and a bit of framing. I find that's one of the most powerful tools you pick up along the journey. It's the five words you say at the beginning of a sentence that change everything. My co-founder actually gave me one of those very early, which is probably why I tuned into this kind of thinking and started collecting these kinds of things.

Wait, what did you mean by the five words you said at the beginning of the sentence?

I'm, again, an engineer. That's always been an interesting experience, with 4 other engineers working for me. I stay pretty current with engineering. I still do it, although not a ton of production engineering. I sometimes start new projects with people, and the first 20 to 30 pull requests or commits in many of the repositories for things that we use come from me.

I like doing this. It's partly fun, and I sometimes need to tap into my craft. Anyway, that's a surprising thing for people, of course. My intuition was always, "Okay, well, this is not the right architecture for this kind of thing, and I know this for a fact because I've built this thing before. So let me go to the whiteboard and just rearchitect this."

Daniel, my co-founder, explained to me, "Well, look, you get pretty defensive then because they just came to show you basically what was objectively their best work, because they presented it to you, and that matters to them." It doesn't matter if you go and make a much better system off the top of your head. If you truly ever do this, they're going to be defensive and try to save face.

So just put "for example" in front of every sentence and say, "What do you mean?" I tried this, and it works perfectly. Instead of just telling them, "Hey, instead of doing this, here's an architecture that you should use," just say, "I could think of doing this a couple of other ways. Maybe it's something you want to consider. For example, what about this?"

And then just do it. Now you're on the same side. You're on a team. We're all working together on the same problem. I'm just adding input. Everything is the same. It's 5 words or so at the beginning. I just find the magic of these 5 words.

Not this specifically—it's things like this. Just frame things slightly differently, and life's easier.

One of the things that Daniel from Shopify—I almost said Spotify—told me, and I want to be fair, was, “You should ask him what were some of the advantages of starting with Silicon Valley?”

I think there’s just too much cross-pollination. You kind of have to sync with companies toward entropy. Entropy creates equilibria. And so, you end up with the companies just being very recognizably similar to each other, and the activation energy for being very different is super high, because that means people come with the wrong ideas into your company, behave incorrectly, and you have to tell them to behave differently.

Whereas, if you are outside of a place like Silicon Valley, where people just move along between companies, first of all, especially early, I traveled to Silicon Valley a few times a year, but I tried to just take it all in and learn as much as I could. Of course, what I got from everyone was not actually the answers to my questions about how to do things. I was getting the aspirations, usually, or the highlight reel.

People wanted to seem smart, right? And so, they gave me the best version of what they thought the answer should be, or what their company ought to be doing. Which was great, because then I took that back home with me—free flights in—and sort of took the ideas that were relevant and tried to always put a Shopify spin on them. I don’t like taking ideas, as you clearly can tell, and never did. I sort of figured out how to localize them to our problems and, ideally, tried to do something better.

Only many years later did I realize that, throughout my career, I was comparing my and our average to everyone else’s aspirations and highlight reel. Just that alone is a really good thing. And if you then try to do better than that, sometimes ignorance is an incredible way of actually just doing better. And so, that helps.

The process of having to translate things, being able to translate things—we had no one. When Shopify IPOed, we had 1 member of the executive team who had ever worked for a public company. That’s it. We had no one who took the company public, and we had no one who was an executive at a public company. We had 1 executive who had worked at a public company once, and it’s just like, I think that’s cool, right? I’ve had a real sense of accomplishment about that because it meant we became our own thing.

Our entire motto in becoming a public company was, “We are not going public. We are creating a public version of Shopify.” And so that was like—

What’s the difference between those 2 sentences?

Again, becoming something else. A lot of people avoid being different when they become a public company because people sort of expect public companies to be a certain way. At least, that’s what people who worked at Shopify did. I felt it was important to be different and do things differently.

One thing we did, and you will appreciate this: we looked at the requirements. We talked about Formula 1. Formula 1 is fun. Sorry, it’s a little bit of a veer, but I love Formula 1 because there’s a rulebook. It’s the opposite of every other rulebook on planet Earth. People read it so that they can comply. The Formula 1 rulebook is read so that you can figure out how to beat it, right?

There’s no engineer in Formula 1 who has anything to do with a central-casting engineer. Central-casting engineers usually want to get requirements and then comply with them. The people going to Formula 1 are the people who want to beat the system, right? This is very important to them. So, I count myself as the archetype of the latter. I would have liked being an engineer in Formula 1. I like beating the systems that other people tell me to abide by.

I heard a story one time that somebody had hacked your compensation system to get more money than they were supposed to. Essentially, they just studied the rulebook, found this loophole, and you were like, “Good for that person.”

Yeah.

And then you’ll fix the loophole or whatever the case is, but that’s exactly—

Yeah.

—the kind of person I want working for me.

Exactly. I mean, yeah, especially if there’s a way. What happens next ends up mattering, right? So, okay, this goes on, and at some point this becomes, “Well, you’re just exploiting a weakness fully.”

I don’t want to give too many provisos, but generally, I never begrudge anyone being an intelligent actor in the local incentive system. Everyone’s allowed. That’s pretty much my morality at this point, or at least a big part of it: you’re allowed to do this. I own the incentive system.

You get too much out of the incentive system, or the incentive system tells you to behave in a way that is not congruent with the company’s goals, that’s my skills issue in designing the compensation. Not you.

Not their fault for—

Yeah.

—taking advantage of it.

But I’d really appreciate it if you gave me a heads-up, because we are still the same team here. Maybe you raise your hand at some point, saying, “This is fine.”

Yeah, not 10 years later.

Yeah. So, back to the IPO: just like in the IPO, we looked for ways—within the constrained rules by the SEC—to see where we had agency and where we could do different things.

We sort of honed in on the IPO roadshow video. It’s a horrible video player, the IPO roadshow website. In every other IPO we could find, there was a CEO standing in front of a screen with a camcorder, just sort of going through the same slides.

Mm-hmm.

We were like, “There’s no rule for that. Let’s shoot a fucking documentary here. Let’s go through the history of why this company exists and why it’s important.” And we said, “What video can be put into this space where no one expects it, that will somehow hook people into full-screening the video within the first 20 seconds and actually get them to watch the entire thing?”

We heard so much about this on the road. So many people were like, “I always looked forward to meeting you. What you guys did with this video was insane. I didn’t even know you could do this.” Every IPO after this was like our way, so it was kind of a nice little moment of putting a bit of agency into a process.

I heard this story one time—I think it was the founder of SiriusXM. I forgot who the person was—but they were like, “Well, how do you make annual reports?” There are no rules around this. You have to present certain information.

Mm-hmm.

So they made a book, and the book was illustrated.

Yeah.

And no one said we couldn’t do it. It’s like, “We’re just going to do it our own way.”

Let me ask you a question, though, because I think I heard you say this as well. You were getting what you felt was bad advice, incomplete advice, or maybe advice that wasn’t from first principles about whether to go public or not. Some people were saying—

But if I said—

“Stay private longer.”

So, yeah, I get you. The advice was always like, “Don’t,” right? Silicon Valley orthodoxy definitely is to avoid it.

I can tell you the story about my decision, but I’m basically predisposed, also, when everyone agrees with something, to take the counterfactual, just to hold it up and see if I potentially like it better.

It’s your personality, an embedded personality trait.

I just find Plan B is usually a better one. Everyone puts Plan B down. I think for many, many, many people on planet Earth, always doing their Plan B would lead to a better experience, right? Plan B gets the ambitious one: “If this ever happens in my life, fuck it, I’m just going to build a company.” It’s such a common Plan B.

That might be what you actually want, but you’re lying to yourself, or you’re scared, or whatever.

Yeah. And also, again, you are a human being; therefore, your firmware comes preloaded with a problem called the sunk-cost fallacy. We just do, right? You’ve got to operate that out of yourself a little bit. You’ve got to be able to discount what has happened more to make good choices.

Because otherwise, you just chase bad with good money all the time. You’re always in the data.

Yeah, I think humans are reflexively—or whatever—just too—

Scared of change, too.

100%.

They’d rather be in a boring—

Yeah.

—or even miserable environment than actually change—

Yes. Yeah.

So wait—

Exactly.

People were telling you to stay private longer?

Being private is better. Friends of mine at Stripe are running this experiment. They managed what they could, in terms of figuring out how to stay private for such a long time, which is actually remarkable. I just think they spent a lot of time figuring out how to stay private. I don’t think that’s the most valuable way to spend time, right? I think you could—

So, you went public in 2015?

2015, yeah.

What were the arguments for staying private longer in 2013 and 2014, when you were having these discussions?

We were tiny. We were 800 people. We were Canadian.

So they were saying, “You’re too small?”

What was our revenue? When Shopify’s first trades happened at, I think, a $1 billion to $1.5 billion valuation, right?

That’s pretty wild—that the public market almost could have gotten venture returns if you—

And people did, yes.

—bought Shopify. You essentially had venture—

Oh, yeah.

—returns in the public market.

Every once in a while, someone puts together a list of the best stock market returns over the last 10 years, or whatever. And, yeah, I’m on that list, right? Just because I decided to take Shopify. I did the old-school thing. That was normal. That’s what everyone did. People went public fast.

By the way, this is kind of important: public markets need to have growth, because where else is prosperity coming from for people who don’t have access to it? These accredited-investor laws and stuff like this, all they do is make it so that when you’re rich, you can invest in private-company stock, and when you’re not, you can’t. Where are the returns?

Of course, this is such a fucking conspiracy to keep everyone private, because that just sequesters all the growth exposure to institutional investors and people who are already accredited. I really dislike that whole thing. Again, it’s about building things and sharing what you’ve figured out, and people buying a ticket to ride on Shopify because they like what we’re building. It’s like, please. That’s clearly what we want.

I just think the institution of public companies is good. My beef is with the whole notion of influence over companies. Again, path dependence. One stock, one vote is like, yeah, but no. Everyone who studies decision-making knows that you want to move toward people who have the most context; they should make the decision. That doesn’t say anything about the leadership of a company.

This is literally what I’m talking about: creating space for creativity, because I want the decisions to be made locally, on the ground.

So, were the arguments for staying private back then just mimetic? Were they simply repeating what everyone else was saying at the time?

Well, no. I think people are generally right. It washes out, right? People have problems with the concepts. Very often, a new idea is dismissed because someone can formulate a reason against it, and somehow that dismisses the entire thing, right?

We live fundamentally—most people spend their entire lives, sadly—in a vetocracy, where you need everyone to agree to do something, but you only need one person to vote against it not to do something. Again, good companies overcome this in important ways. Similarly, one good argument dismisses the entire idea, which is a very common way to evaluate options.

What people very, very often don’t do is reformulate all the arguments against the status quo, right? The thing you’re doing right now is also flawed. It’s just that we really understand the upsides and downsides. The new thing might have way higher upsides, or actually, maybe not. Maybe the upsides are obscure or hard to know for sure. But the downsides are clear, and they can be formulated; therefore, the usual default position is not to do it.

That, I think, is incredibly undisciplined, because you need to think in propensities and chances. Sometimes you need to really load up a decision with all the pros and cons that can be articulated and, in your mind, look at where the scale comes out.

For a company that’s trying to sell an e-commerce platform to everyone, especially for many American companies, being a Canadian private company that’s tiny is not ideal. People don’t usually go public because of marketing or name recognition, but it’s—

Super valuable. Anyway—

Put it on the pro side, and then put everything else on the pro side, too. We might get better people working for us, because some people really want liquid stock, and so on.

In the end, it’s not even close. The downsides have been a couple of annoying phone calls I have to do every couple of quarters, at least for the first 10 years, I suppose. I thought that most of the process was actually quite valuable. I learned a lot about my own business, having to—

Explain to others.

Exactly.

There have been a few previous guests on the show—Brad Jacobs, John Mackey, and Daniel Ek, all obviously running public companies. In Brad Jacobs’s and John Mackey’s cases, they both said—Jacobs makes the point that, one, he thinks it’s wonderful free marketing and advertising; two, you now have a currency.

Mackey said the same thing. It benefited Whole Foods to be public fast, because then he used that currency to acquire a bunch of his other competitors and accelerated his growth.

That’s such a good point.

But Brad Jacobs’s point was that he likes the pressure of that check-in—

Yeah.

—where he’s like, “This is what I’m trying to do with the company. This is what I’m saying I’m doing with the company, and these are our results.”

Mm-hmm.

“And I want somebody to say, ‘There’s a big disconnect here.’”

Mm-hmm.

“And I want the pressure to actually perform against some of these expectations externally.”

We talked about the environment that you’re building at Shopify. I’m curious about how you think about the talent that’s inside the environment.

Yeah, of course. I mean, arguably, I really, really believe in the Stoic, maybe Adlerian idea of separation of tasks: What do I have control over, and what don’t I?

I always love the particular formulation about thinking about how other people think about you. Again, you need to make yourself a person worthy of being liked. If someone then dislikes you, that’s up to them. It’s not your job to make people like you; it’s your job to be likable. If they make the category mistake, it’s their own fucking problem, right?

It’s a really, really freeing way to think about talent. My job is to make a company worthy of the best and brightest to work for. This is the part everyone skips. Everyone’s like, “Oh, we need to hire a better recruiting team.” Well, maybe you also need to be worthy of the kind of talent that you would like to have.

This goes from office space, if you have any, to the company—the amount of bullshit people have to deal with. How much bullshit and politics can you subtract from the whole thing? Can your large company feel like a small company with lots of resources and lots of impact?

It’s more fun to ship software to millions of people that you really respect than to a few people who you then try to make a sale for at some point. Impact matters.

In a way, this is likely going to be misquoted, but I think the talent eventually takes care of itself. There aren’t that many good companies to work for. There aren’t many good companies that deserve the attention of the most capable people, because they have the option to start their own companies.

That’s a very interesting idea. There actually aren’t that many great companies. It’s your job to make your company one of those companies that the best people in the world will want to work for.

Wait, what do these people look like, though?

The reason why I didn’t move to Silicon Valley is because I knew how to create geographic consensus on the Eastern Seaboard that Shopify was the best company to work for, full stop.

And I just knew—

How?

I knew I had the ability to do it. Given that I estimated I would need somewhere around 1,000 people, or something like this, back in those days, I thought, “I’m going to find another 1,000 people who think like me about how good a company needs to be to work for, and I can dislodge all of them.”

If I go to Silicon Valley, I can’t pay what Google pays, potentially, or someone else makes up for their skills issue by just overcompensating. It’s called compensation because you kind of have to compensate people for the shit you ask them to do. If there’s a lot of shit, you have to compensate people for it. But if you have an enormous amount of resources, you can make up for skills issues with money.

The East Coast was much more hedge-fund-y and much more high finance, and for the kinds of hackers I wanted, I knew how to do this.

But wait, I want to pause there. I love this idea, where you’re like, “I’m going to build a company that I would want to work for.”

Oh, yeah.

That’s on one end of the spectrum, which is great. You also had this thought experiment that I heard you say before, which is that one of the worst things that could happen on the other end of the spectrum is that you wake up 10 years from now and you’ve built a company you don’t want to work at.

Yeah, totally.

And then you have this great line I want to pull. It’s like, “That’s dystopian. There are so many companies I wouldn’t want to work for. Why did I use my life to build another company I wouldn’t want to work for?” You should just build a company that’s worth working for.

Yeah. Again, you should take pride in your work—in the inputs and how you accomplish the things you set out to accomplish in the aggregate, right? Be easy on the path, but the outcome should matter to you.

I shake my head sometimes at how people overcomplicate these things. People know what we’re looking for, right? I’m talking about very, very, very fortunate people here—people who have a lot of choices, who have incredible competency and some of the rarer skills, combined with enormous drive.

There are a couple of different attributes that all have to come together to be one of those people who can command hundreds of thousands or millions of dollars in compensation packages that you can make in this industry.

Partly because you're worth it, right? You're contributing more than that to a mission like this, and therefore you can command this kind of compensation. That's the way this all fits together. I think of everyone as an entrepreneur, really. Your capability is that you are the entrepreneur of your own work output.

Your product is what you can do and what you can accomplish. Your market is the companies that exist, and you are selling an exclusive subscription. That's the institution of employment, right? It's a two-sided opt-in relationship. Whenever people come to me and say, “Hey, how do I earn more money?” I'm like, “Well, be too good to ignore, then we increase compensation.” You know, be better.

You're making this amount and would like to make this amount. In which way have you become so good that you can add that amount of increase to the collaborative project we're all working on? I think this just clarifies and simplifies things, right? All of this said, I don't want to make it sound like it's inevitable that if you just do all the right things, you end up there.

There's always an information gap and a reputation lag, and I'm out there recruiting and telling people about the company a lot. What helps me a lot is that people are curious about the company and have heard about the company by now. But that is because of things that were under my control. I tell people that it's an unshared-attention company. There are no part-time jobs.

You will have to put a lot into it, and you're going to get even more out of it. You will make great money, but we attempt to never be the highest-compensation play. We have found that the people who optimize for that above all else actually don't do well at the company.

Because Shopify is a mission-driven company.

Exactly. In terms of your own skill set, partly by demand, partly by osmosis, and partly based on the culture that's surrounding you, you will advance vastly faster at Shopify—building capabilities, skills, and correct ways of thinking and mental models and these kinds of things—than, I think, at any other place.

Therefore, the compound return to your career is going to be vastly higher here than anywhere else, because you get to a potentially higher compensation level earlier in your career because you're worth it. The entrepreneurial story is very fractal in Shopify. It's just like, what is your product? What's the market, and does it fit? It's a clarifier that I think is worth talking about.

I'm going to quote David Ogilvy again, because I think it's interesting. He says, “Talent is most likely to be found amongst nonconformist dissenters and rebels.” When I asked the question, “Hey, we know the environment. We have a description of the environment. What are the people in the environment?” there were 2 ideas I got from 2 separate entrepreneurs.

I've mentioned both of them already because they used the same terminology when they were looking for talented people to hire in their companies: Daniel from Spotify and Karim from Ramp. They both said the same thing to me: “They hire for spikes.”

Mm-hmm.

Spikiness. Yeah.

Spikiness. So, a lot of big corporations want you to be well-rounded. Like, no, no, this guy might be a bipolar alcoholic—

Exactly.

—or this guy might have something else wrong with him. He might not shower.

Yeah, yeah.

Like Steve Jobs—he was hired at Atari. Everybody was like, “Nolan, Nolan Bushnell, the founder of Atari, you can't hire this guy. He stinks, and he wants to sleep at the office.” Nolan's like, “Yeah, but he's really talented. So he's going to sleep at the office, and you just have to hold your nose when he's around.” Do you agree with the idea?

100%.

Okay. We never look at the credentials and stuff like this. Again, I'm a high school dropout. I would be rich if I would start making everyone want to have a PhD. The path by which we do it is that we go through a life story.

We ask you, “Hey, give us your story.” It's a very important ingredient in our hiring process. Frankly, what we're looking for is high-agency experience. We look at it like, “Something went wrong in your story at some point. Well, now let's zoom in. How did you react? Give me the minute-by-minute.” This is what we want to know, right?

A lot of us are entrepreneurs. I talked about the founders of companies we purchased, but most of the people in Shopify, just in general, are entrepreneurs. Many started Shopify stores, built a business, and then fell in love with the company, applied for a job, and got it. So there's a huge amount of company-building background, and we want to be surrounded by people we admire.

That's, I think, the biggest product any company can deliver to its employees: that every day they are surrounded by people they deeply admire and can learn from. That's a very important part of this. So, lots of entrepreneurship, lots of entrepreneurial impulses.

Then we look for this high-agency behavior and, obviously, did you drive a particular craft that you're applying for to excellence? How do you think about this kind of thing? Can you talk about excellence, or actually perform it? This is really what we fill our buildings with.

Again, we create the space and let them be creative. This is how it always worked, and it's worked really, really well. We couldn't do anything other than this in the beginning, right? In the beginning, we had no money—none, like, really zero. I didn't take a salary for 4 years.

I heard your father-in-law covered your payroll when you couldn't make it.

That's right. We were living with my parents-in-law, in my wife Fiona's old bedroom. I put an IKEA desk in it, and I built a lot of Shopify from there. I just kept costs low. That was the main thing.

Making every dollar count is something I'm so glad we learned. This is what I worry about sometimes with all these really, really big seed rounds. The main thing money does is it gets you a whole lot more of what you had before. Sometimes it reveals things, but usually it just gets you a whole lot more of what you had before.

People did raise early partly because they were not tight with money, and you get a whole lot more of that not-tight-with-money-ness at larger scale. It gets really, really bad.

You're making my heart sing, because I feel like I'm on an island by myself about this. I've spent 10 years reading 410 of these biographies. All of them control their costs like crazy. They'll tell you that constraints are your best friend.

I usually don't meet startup founders because I'm not an investor, but a good friend of mine will say, “Hey, meet this young kid.” I met him, and one of the things he was talking about was, “Yeah, let me show you pictures of this beautiful office we're moving into.” He was showing me everything, and I was like, “This is stupid. You haven't done anything. You don't deserve this.”

Go look at where Apple started. Look at Microsoft. Microsoft started working in a strip mall in Albuquerque, New Mexico.

You need a triplex in SoHo? What are you talking about? The question that I posed to him was, “Okay, how does this make your product better for your customers?” His answer was, “Well, I'll be able to recruit better talent because the office is nicer.” Then you don't have a mission. It can work, but I feel like you're not increasing the—

You built a $200-billion company in your wife's childhood bedroom. That's where you started. Maybe we should repeat that over and over again. Just go look at where these companies were.

You mentioned the Larry Ellison episode I did earlier. They were in a shitty office building.

Yep. Yeah, yeah.

I think they had 3 separate rooms. Oracle—that's where Oracle started.

I think there's a well-documented effect that when companies build monuments to themselves, they usually decline right after.

Yeah.

That has a long-running history in business. I do think there's an argument that companies underinvested in office space for a very long time, and it actually was a mistake. I think the space around us is extremely important, but it doesn't need to be expensive. That's probably an aside, but—

Oh, let's go down there. What does that mean?

Again, we didn't have a lot of money, but we wanted nice office spaces. This was the decade when there was a lot of debate between open-concept and private-office spaces. It was sort of a fight.

My co-founder is Daniel Weinand. He's a polymath. He learned programming when I asked him for some help with programming, and he was immediately good at it, which was annoying as hell. Before that, he was our designer and literally everything else, too. Now he's a DJ, and before that he won poker tournaments.

It's really, really helpful to have someone like Daniel as a childhood friend because it really, really misleads you about human capabilities. If this is just your buddy who does stuff like this, it changes your understanding of what's possible.

At some point, he really tapped into this idea: no one can be more creative than the space around them, and no one can care more than the person they work for. That's the way he created this philosophy around it.

I told him, “Look, here's the maximum budget, and we're getting B-grade office space. That's kind of cool, and you need to figure out how to create places that are a 10 out of 10 for doing world-class work.”

But inexpensively.

But you need to figure out how to make this cheap.

Man, we built a bunch of offices in a row, and Daniel was insanely good at this.

Yeah. How many years into Shopify?

This is like 2010, I think. 2010 was the first one. It was the 6th year since founding, 4 years after Shopify launched. We went from being above a coffee shop, which was really nothing, to our first office.

So how do you do this?

Daniel, like I said, first of all, the dichotomy is wrong. It’s not between open-office and closed-office spaces. There’s a perfect midpoint, and it has to come from your strategy of products.

Shopify loves the 5-person team. We increase to 8 sometimes, but we think the best team size is 1, because a single author can do things that are impossible for teams and hit high notes that are unreachable. But I think product has barely explored the field of what can be done as solo projects. Actually, it’s changing again because of AI, which is an exciting bit.

Most projects worth doing need to be done in teams. Then there’s a magic number at 5. It’s sort of what the military ends up figuring out, too. They test these things and come to the same conclusions. You can temporarily go up, but then, at some point, you have to split teams and parcel out the tasks. Each of these gradations is, I think, a 10× loss of productivity.

Shopify is like—our R&D team is, I think, 3,500 people. That’s really lots and lots and lots of small teams. This is also why we need the legibility and coordination layer that I talked about earlier.

Okay, so pods. Daniel figured out how to make pods that are exactly right. They are private enough while being open enough, and never make the office feel dead, but they also throw you around.

What’s a pod? What does that mean?

It’s a room that seats up to 6. When you put the 7th person in, it’s uncomfortable. Again, think about how much policy you don’t have to post if the environment just makes you do the obvious thing. You can plainly feel when there are too many people in this area. When you know what you’re doing, you don’t need to post a policy about how many people you want in a pod.

People self-organize into 5-person teams because that’s the most comfortable thing to put in a pod. They’re not meeting rooms. Meeting rooms go in the middle on our limited floor space, because we don’t want people to spend all day in meetings and then get out of the dark rooms.

Everyone sits around the windows in pods, and they’re structured in such a way that it all works. The coffee area is on the other side. To create serendipity, we eventually cut through every floor to create stairs. We moved teams in such a way that certain teams that had to work together but worked on infrastructure were separated across floors, so that, going to each other, they would run into other people who were using the infrastructure, too, and so on and so on.

So, design: being super conscientious about this is really important. Daniel did an amazing job. I think we open-sourced the floor plans, and I think lots of this work has gotten absorbed into the tech industry as the right midpoint. This is not interesting anymore when you look at what they look like, but many people cargo-cult it. They just Xerox-copy it, essentially, instead of understanding it, so they make pods too big or whatever. These kinds of things.

We’ve talked about cargo-culting a few times. You observe that behavior in other companies. I’m curious what, internally, you have caught your own company cargo-culting.

Oh, we have a Toronto office here. The Toronto office actually has a few floors that are clearly a mimicry of these ideas that they figured out, but they don’t understand where these come from. Again, the pod’s too big. The nice meeting room is right on the window and has lots of light. Maybe this is okay to have a few, but I went through it with the teams because we then changed some of the floor plans, and they moved here and realized, like, “Holy shit, this is like…”

Norman doors everywhere. Fuck! That’s the craziest thing.

Wait, what?

I’m fundamentally allergic to Norman doors.

What’s that?

Doors that you push, and then—

You can’t tell.

You realize you can’t tell.

Yes!

Yeah. This is the most fundamental design error that you can possibly make: creating affordances in a product that make people predictably do the wrong thing. Literally everyone who’s ever been involved in the process of designing a Norman door should have their design license revoked. I’m being facetious, obviously; that isn’t a thing. But it’s just like, man, skill issues with doors are a very common occurrence, I think, in places that are not conscientiously designed, and all of those doors are in place very fast.

The funny thing is, you can usually turn a Norman door into a non-Norman door by cutting things away and putting a push plate somewhere, so that helps. I make a big deal out of these kinds of things. I cannot surround people with things that are underperforming the quality of engineering we want to do.

And that’s down to the detail of the doors in the office?

Yes.

Wow.

Yeah. Yeah, of course. Product is not a thing. Product is an abstraction. Product is a name, like a handle that you put on some totality of a body of work. From the perspective of people creating a product, the product doesn’t exist. It’s what other people call your work.

A product is just details. It’s many, many little details that are appreciated if they compose right and make a bloody mess if they don’t, right? From the inside perspective, you have to sweat all the details. The way you do anything is how you do everything. That is absolutely down to the floor plans, the ambient noise levels, and so on.

You go in a room and it’s silent. You notice when you’re talking while going to a meeting room, at least on the floors we fixed based on specs. It’s kind of uncomfortable to talk in the hallway because it’s so echoey. Then you walk into a meeting room, and immediately it’s just a different noise level because it’s diffused.

We have enormous amounts of—In fact, we actually worked with companies on Shopify to create sound-dampening products that now supply all the tech companies, which is really fun. In fact, everything in our offices is from Shopify stores, right? Because, again, if anything we are doing is not mission-aligned, we just didn’t align to a mission, axiomatically.

You can go pretty far in these kinds of things. When you need something for office spaces and it’s not on Shopify, we usually put a call out for someone to start that thing so it’s on Shopify, and we help them get started and build the thing. Then we can maintain our 100% Shopify quota.

That’s incredible.

We did at least one enterprise sale with a supplier because we really wanted their floorboards for something. They really looked great and were at the right price level, but we can’t place the order unless you come to our price. So sometimes we’re maybe cheating.

That’s actually a good way to get a customer.

Yeah, it’s also a funny way of doing enterprise sales.

I want this thing, but you just need to migrate over to our platform first.

Yeah.

All right, I want to talk about video games. I want to talk about AI. I want to talk about them being related. I mentioned my friend Karim a few times. I went to his house last week, and I was like, “Dude, I’m talking to Tobi. He’s also obsessed. He has an engineering mindset like you. He’s obsessed with playing video games.”

And you’ve mentioned a few times where you thought you learned more about building a company from StarCraft than by talking to a bunch of people or reading a bunch of books. But then his whole thing was just that building with AI now has to feel even more like a video game.

Yeah, I’ve said Factorio, yes.

Okay, so that’s the example he gave me. I haven’t played the game. Can you elaborate on this?

I completely agree. Again, the world has such a weird way of somehow changing itself around me to make the stuff that I thought was irrelevant relevant so late in life.

It's absolutely amazing. Video games are best for this, although it depends very much on the video game. In the aggregate, very often, the good video games are simulations. They are a world unto themselves, and you are a high-agency actor: you modify things, perform actions, and make decisions. Then the game rearranges itself, usually giving you a very quick feedback loop because otherwise it's unfun. You learn about the consequences of your actions.

Competitive multiplayer games are very obvious examples of that. StarCraft and strategy games aren't that popular as a category anymore, but when I was going through my teens in the ’90s, StarCraft was the biggest game. I loved it because it was really, really hard. It was easy to learn, hard to master, which is exactly, I think, a hallmark of everything that's worth doing: motorsports, kiteboarding, my hobbies, tennis—all these things are like this.

Specifically, StarCraft is honest in a way because there are 2 people on the map, mirrored. Everyone has the same potential in every game, and 1 person wins at the end, right? Additionally, it's a game of imperfect information. It's not like chess; you don't see the moves. You see only some information, and, in fact, you have to invest. You have to use one of the things that could get you resources to go to the enemy base to figure out what he's building, and then you need to interpret this.

All these kinds of things mean that what you learn in this game is that information is everything. There's no right decision. There's only the context in which decisions turn out to be correct. While resource management is extremely important, you have to understand that resource management isn't just quantifiable. It's not just minerals and vespene gas that you have to manage, which allows you to build units. It's also your attention. Frankly, attacking other people's attention is much more profitable, very often, in a game of StarCraft than actually attacking their base. If you overtax their ability to pay attention to things, they will make mistakes.

How do you overtax their ability to pay attention to things?

You can do an early attack. You can do any number of things, especially if you play against the same person. When you get to a very high level, you'll play the same people over and over again. They get a sense for how you play.

At that moment, there have always been people who are incredibly prepared. They've perfected playing a single way, whereas my play was always different. Chess, StarCraft—it was all the same. I disrupt the other person's play. I do moves that aren't in books. They're not in books because they're not good, but other people can't deal with them because they need their preparation.

Their ability to play chess is crystallized intelligence, not fluid intelligence—the ability to roll with the punches. I learned a lot about myself doing this, but I also learned a lot about how to approach a game. How do you get better? How do you get more out of every game you play?

I never, ever in my life had as much time as some other people had to invest in something like this. I always had to ask myself, "How can I get more skill-progress units out of the time I have than other people get out of less-constrained calendars?" I'm talking very fondly of StarCraft. It might just have been oddly the right teacher for me at that moment, and the student was ready. I could have maybe learned the same lessons from any number of things.

I think it was specifically good that it was a simulation. You started a new game, and it had skill-based matchmaking, so you should play against other people of your skill level, based on Elo rating. You had a very clear sense that if you started winning more games than you lost, that meant you were progressing. If you started losing more games, you were regressing on an absolute scale compared to all the other players in the world.

I think that was a perfect little sandbox to explore how to think about when it's time to build infrastructure, when it's time to invest in resources, when it's time to prepare, when it's time to reveal your hand, and when it's not time to reveal your hand. I think it was a perfect place for me to spend time in my teenage years, given what I did afterward.

So, back to your AI question: I find that working with AI agents—especially now, as we're recording this in week 3 of 2026—is fascinating. If we had recorded this in week 52 of 2025, it would have been different. That's how much is happening. The world changes every 3 weeks right now.

We've figured out some new harnesses for agentic loops that are much more autonomous and, frankly, supported by models that didn't exist at the beginning of December, which can perform at this level. I look at my computer, and I have 6 different agents going, all coordinating among themselves. They send emails to each other, which I think is hilarious.

I'm thinking, "Man, this is starting to really look like StarCraft." My attention is divided: this one is currently working on a module that I really, really need, so I'm zooming in, paying attention, doing a little bit of micro, and giving the output of the critic a little extra double-check. The critic looks at all of them all the time to figure out whether they're still on task. It's the exact same thing.

It is a fascinating time. I just can't describe how fast progress is right now. You know what I tell my team? For 21 years now, starting companies has been really, really, really hard, right? The first years are interesting, and everything's urgent because you're getting to a margin. We got within a week of running out of money so many times, and sometimes we were bailed out.

Many times, clearing the one-time hurdles I've gone through came down to millimeters. I bet that if you reran a simulation of the first 6 years 10,000 times, it wouldn't succeed in most of them, right? In a way, luck is a big component: timing, all these kinds of things. People's luck, especially.

After that, the IPO—you'd think that's probably the most interesting period of a company. I think it was hard and maybe not terribly interesting. It's a financing event at the end of the day. Maybe some aspects were interesting, like figuring out how we made it our own as a process. Partly, this is why I did it. This is why we made changes to the process, or why we wanted to have our own stamp on it, because I didn't want to join someone else's process. I wanted to make it my own.

Then there was COVID, which was very, very hard—definitely not interesting, actually fucking annoying, mostly. So I'm trying to figure out 2026. This is going to be hard and interesting.

Those are the things you're most attracted to: hard and interesting?

Year 21 is going to be the most interesting year in this company's history, probably—literally, I think—in everyone's career. Everything is lined up. It doesn't matter what you've got; everyone's going to be measured against how well you prepared yourself. What's your skill set? How quickly can you react to everything that's happening? How quickly can you readjust? How quickly can you rederive everything?

You put out this internal memo where you're saying—I think you know the language better than I do—that it's basically a company expectation that the first thing you do when you have a problem is reflexively try to use AI.

Yeah.

When did you put that out, a year ago?

Yeah, no, maybe not quite.

But something along those lines.

And people were just like—

Yeah.

"This is ridiculous." There was some pushback.

Yeah.

And now, to your point about how fast everything is moving, it's like, of—

Yeah.

—course, you would.

Yeah, exactly. People will look at that memo in 2 years and say, "He didn't say anything."

Right. It's like saying the sky is blue.

Yeah, exactly. So does this excite you? What's it like?

Oh, yeah. Man, I love that. It's like this is what I'm here for. If AI hadn't happened, I don't think I would run Shopify anymore. There are much better leaders when times are stable.

Whoa, what?

If times are stable.

Yeah, but I thought you were in love with this problem, this unsolvable problem. So you'd still work on it, but you wouldn't be running it? What do you mean?

I mean, I think I would—no, I might try my hand at a different aspect of helping people self-actualize. My personal mission is to create more entrepreneurship. I'm going to judge my own career by this: the thing I discovered that was meaningful to me, have I helped the maximum amount of other people do this thing?

When did you realize that was your mission?

I know exactly when, because it was while I was preparing for one of those internal events, the summit talks, which are always a bit of a soul-searching exercise. I do a lot of this message-in-a-bottle reading and self-introspection. It was 2014. I was preparing for my summit talk, and my biggest struggle was with my own identity.

I think I was mentally preparing for the CEO cosplay period, which I wish I hadn't. But I like engineering. I want to spend my day hacking on stuff and building things, not managing. Again, I used my trick partly to give a talk on something that I had come to realize but maybe hadn't fully internalized.

Basically, I said I loved engineering because it turned out that was how I, as a kid, could build things when I started doing it.

What I actually love doing is building things. Specifically, I love building things that share something that I have experienced. This is a moment of becoming an entrepreneur: getting your first sale. So I made a talk about combining those 2 things and saying, “My mission in life, my craft, is actually building things in general.”

I build. I started with products, like code. I then create products, and now I create companies that create products, and I engineer those. And that causes more of this kind of thing.

So I kind of wrote that as a personal mission for me and said, “I will attempt to run Shopify and try to be the best CEO I can be until someone taps me on the shoulder and says, ‘Other people could be better for this company,’” as long as it’s in cohesion with my personal mission. But I’m also perfectly fatalistic about this kind of thing. I don’t do the job for money. I do it because I feel like I can contribute a lot.

I love this idea that you view entrepreneurship as a way to get self-actualized.

Oh, yeah. I mean, what’s more honest?

But it’s also your form of entrepreneurship. I was reading—I never read comments on podcasts and YouTube videos. I highly recommend against that. But people were saying, “This is one of the good entrepreneurs.” What you’re building, and the other entrepreneurs you’re trying to enable, it’s not like they’re crony capitalists. There are a lot of dirty parts to it. I’m as capitalist as they come, so I understood what that comment meant.

Out of every single note that I took from all the talks that I’ve seen you give before, the one at the top of my list is exactly how I feel about what I’m trying to do. You said, “Focus on craft and giving a shit.” And you said, “It doesn’t matter that a bunch of this stuff isn’t quantifiable.”

It’s such a basic thing, but why is it so simple? It’s so simple, but it’s just not easy, right? In any which way. Companies hate the unquantifiables, in fact. It’s amazing to see it.

You hire the first team together, and if you get to talk about building a company on a podcast, that means you were successful. So the first team clearly did this, and therefore, the first team was very good. You hired the first team based on your gut. You just knew these were the people you needed. You make mistakes, but you correct. And that’s all just judgment, taste, maybe even bias.

I think we need to move nostalgia into the bad category of words and bias into the good category of words, because bias is actually perfectly fine. Without bias, you actually have no convictions. You should trust your convictions. This is really what I learned from the thing.

You make the first team as a pure play: this is the best I know how to make decisions. And then everyone wants you to write it down. You know what happens when you write it down? “Here are the things that you are looking for.” The people who read it are the people who are very, very good at following lists. The people that you probably described are not the people who go look for a cheat sheet.

The people you’re trying to avoid hiring are the people who would study a cheat sheet very well. So you just gave everyone a way of performatively doing the thing that you want to see without being intrinsically like this. And so you end up very quickly losing your company once you systematize things. However, everyone is telling you to do it.

There’s a giant conspiracy against these things: taste and quality. If it can’t be put in numbers or written as a list, companies hate it. Yet almost everything you have to do to build a great company involves these things, and you shouldn’t write them down. Actively avoiding writing something down is sometimes the most important thing you can possibly do.

Do you think it’s also related to the way you brought up sunk costs a bunch of times?

Yeah.

Where it’s like, “Well, I wrote this down. I might change my mind, but there’s some kind of human element where it’s like, ‘Oh, I don’t want to contradict this because it’s in writing.’”

Yeah, exactly. Again, people like to be consistent over time. I just don’t find that important. I found the cheat code to always being right is just to change your opinion every time you get better information. Because as you get closer to a point where you have to make a decision, that means you will wind up with the right one.

If you don’t stop at some point to incorporate new information, you’re going to have a problem. And even if you are wrong, and if you figure it out again, what do you do? You have one momentary chance to steer your concern toward the best possible outcome. Are you going to take it, or are you going to be the person who just leans into the comfort of some kind of character consistency? Why? For what?

Again, what’s the job you want to do? You want to look good or be good? It needs to be okay to change your mind when better information comes along.

And again, here’s the crazy thing: I think if you go through the transcript of this podcast, you will find—or just ask an AI to do it, and it will probably point out—various contradictions inside of what I’m saying about what to do. I’m in this. I’m probably on multiple sides.

The reason is actually—this is almost the most important thing: everything comes in layers. There are so many different levels. What’s the optimal operating system in some frame of reference is often contradictory to, and the opposite of, a larger frame of reference.

I like this box, talking about boxes, because inside of a box, after a while, you can have a map of it, illuminate all the corners, and understand what’s there. You may decide that it wasn’t the right box to pursue. Maybe then you need to figure out—maybe you need to bend it and say, “Okay, actually, the problem is an aspect of another problem. Maybe if you put it in a larger box and solve that, then this problem doesn’t even need to be solved.” That happens.

But this is exactly what I mean. As you zoom out, you actually need to completely change the way you problem-solve, because you’re starting to work much more with longer timelines. You need to make bets at a high level, but you shouldn’t make bets with infrastructure, because you can.

In engineering, again, you can run benchmarks, figure out if something is fast enough, and it’ll be a yes or a no. You don’t need to bet on the trajectory of databases growing faster at a certain thing. You don’t need to. You can measure. So there, you want to be super-quantifiable.

When you try to figure out which market to go into, you need to figure out, well, right now, how AI is going to change that market. You need to figure out whether, if you join a market, that market is going to be a temporary, unnecessary aspect of something else we are already doing, and whether it will be absorbed into a different category of products. Is this still valuable for you to do, and so on? And what are humanoid robotics going to do to the world of logistics? You’ve got to work with trend lines mostly, right? That’s not important in many other kinds of decision-making.

So I think there are probably 4 or 5 different levels at which you have to look at a problem. That is actually a lot of the skill of a good executive: being able to meet the team where they are and help them talk about the outside world, quote-unquote. Here is how their work relates to all the other frames of reference, all the other abstractions, and how it sits as a part of the larger company.

Everyone should be able to connect their day-to-day work back to the mission. You work on customer support? Well, you’re helping people literally not give up on being entrepreneurs. It’s not even subtle. It’s actually an easy exercise to do when you ask people to do it.

But it’s distressing how many companies leave it unstated. The company mission is some kind of set of attitudes or something that isn’t actionable. Those are unforced errors, and I think things go better when you get them right.

I love the idea of tying everything back to the mission.

I do hope people go through the transcripts of this conversation.

Mm-hmm.

I think it’s going to be incredible.

I printed out some of them, and I’m sure people are going to tell me about all the contradictions.

And I think, again, this is the fun bit. I find contradictions in my belief systems all the time because we are all balls of contradictions.

I’ll get messages. I’m like, “I’m listening to this episode you did in 2019, and you said this, and on the new one, you said this.” I’m like, “I’ve read 250 of these things since then.”

What do you think I’m doing this to stay the same?

Right.

What’s the point?

It’s a good point.

I had high expectations of this conversation, Tobi, and you exceeded them. I hope we do this at least once a year. I think the next time we do this, we design with the constraints you like, and we just evangelize entrepreneurship and the fact that it’s one of the most important things in the world. I think you’re creating a lot more entrepreneurs, and I’m trying to create a lot more entrepreneurs with my books as well.

I’ll tell you one thing. I’m not sure if I’ve shared this publicly, but I think you are going to get more mileage out of this than anyone else. When we did surveys with our customers because we needed to identify where our customers were, what they did, what they read, and these kinds of things, what we found—this was a long time ago, before your podcast existed—was that almost 80%, or a high 70%, of people checked the box saying they had someone they could send an entrepreneurial question to who would respond to the email within 24 hours. I initially thought that was insanely cool. How many people have such a resource?

It’s not like that in the general population, though, right?

Eventually, I clued in. I think it was probably on Wikipedia when I read about survivorship bias, and I realized, “F***, the reason they are my customers is because they have someone who responds to these questions within 24 hours.” It’s the other way around. It’s causal.

The concept of entrepreneurship is so underexposed and underexplained. It’s gotten better in the 10 or 15 years since those surveys, but I love what you’re doing because you’re rolling this out for people to introspect—an entire life story of entrepreneurship—giving people the thing that is implied by the question. It’s like, do you have any exposure to the world of entrepreneurship in some way, so that you actually know what you’re signing up for? Because the people who don’t know it exists, the people who have no relation to it, or the people who don’t happen to have someone in their family or among their friends who has, at some point, started anything, well, they just won’t try. It’s not part of their toolbox.

Just putting this into more people’s toolboxes is going to be so load-bearing. I think, actually, in this world of AI, there’s going to be disruption, and so many more people than people believe will have entrepreneurship as a plan B. Making things will be much simpler in the world of just-in-time manufacturing, 3D printing, additive and resin printing, and all these kinds of products—humanoid robotics, drones, lights-out factories making anything on demand, and 3D printers in every home, or at least very, very commonly in homes now.

In that world, more people will realize that working with AI is about learning the skills of figuring out how to make a better cup, how to design a better version, and how not to have a Norman door. It’s capitalism—the best parts—and I think it will be much more common and much more exciting. It’s so fun to work on.

In that way, you and I work on the same project, which I think is super, super cool. That’s because the information gap is about getting the stories out, telling people that’s possible, and it’s so important for people to then try. Hopefully, they might use Shopify, and then hopefully we can push them forward.

I love that. I appreciate you saying it. I appreciate the time. Thank you so much, Tobi.

Awesome, man.