Safe enough to try: An interview with Zappos CEO Tony Hsieh

Organizations are more likely to innovate and thrive when they unleash the potential of individuals and the power of self-organizing teams, says the online retailer’s CEO.

Tony Hsieh, the CEO of Zappos for more than 17 years, is not afraid to create “a little weirdness.” In fact, that is among Zappos’s core values. The company that got its start selling shoes online, became known for its near-fanatical devotion from customers, and was acquired in 2009 by Amazon has more recently been pushing the envelope in another area: its organization. Four years ago, Zappos kicked off its high-profile adoption of holacracy, an organizational model that distributes decision-making authority in self-organizing circles, made up of employees who hold roles (often more than one at a time) rather than job descriptions, with each circle arranged around a purpose statement. These experimental approaches, Hsieh hopes, will enable every employee to act as a “human sensor,” and the organization as a whole to be more adaptable, innovative, and resilient. While Hsieh doesn’t claim Zappos is an easily emulated model, the company has become a thought-provoking test bed for organizational ideas whose ultimate impact will become clearer in the future.

In May 2017, Hsieh sat down with McKinsey senior partners Aaron De Smet and Chris Gagnon to share his views on organizational values, purpose, and decision making; the importance of the individual; and the potential for self-organization to generate innovation. The interview, which took place on the Zappos corporate campus, in Las Vegas, Nevada, gave Hsieh an opportunity to get beyond the headlines as he described the philosophy behind Zappos’s evolving organization.

The QuarterlyWhat does holacracy mean for Zappos?

Tony Hsieh: People can get caught up a little too much in the technical details of what holacracy is or what tools we’re using. We’ve always encouraged employees to move around to find the intersection of what they are passionate about, what they are good at, and what adds value to the company, even in the “old days.” For me personally, calling it “holacracy” was more a way of codifying or making explicit what was already implicit in our culture. We shouldn’t have to be dependent on a benevolent manager or CEO to allow employees to move around within the organization, because that’s a single point of failure. Our org chart is available in real-time online and changes probably 50 times a day, and every one of our 1,500 employees can transparently view what every employee’s purposes and accountabilities are. We have self-organized governance methods and meetings that happen on a regular basis, and it’s all browsable and updateable online, along with, occasionally, policy updates—all of which enables any employee to contribute to the evolving structure of the organization. So it’s not so much about “holacracy” as it is about “self-organization.”

Video
Teaching holacracy

The Quarterly: If a company is self-organizing, and being dependent on a CEO can be considered a point of failure, how does the company keep its bearings?

Tony Hsieh: Imagine a greenhouse with lots of plants, and each plant represents an employee. Maybe at a typical company, the CEO is the tallest, strongest plant that the other plants aspire to one day become. That’s not how I think of my role. Instead, I think of my role as the architect of the greenhouse, and to help figure out the right conditions within the greenhouse to enable all of the other plants to flourish and thrive.

Cities are another example of self-organization. Cities are the man-made organizations that have best stood the test of time. Cities last much longer than companies. Cities are resilient. Cities are adaptable. And cities aren’t hierarchical the way most companies are. I read somewhere that all of Manhattan has literally three days of food supply. But there’s no central food planner for Manhattan. Instead, you’ve got consumers and businesses “selfishly” consuming food in a self-organized manner, which creates opportunities for suppliers and so on. And that self-organized system works if there is a natural disaster; a bridge can go out and Manhattan still doesn’t run out of food.

Not only do cities stand the test of time, there’s plenty of evidence they actually scale in terms of productivity and innovation. One interesting statistic is that whenever the size of a city doubles, innovation or productivity per resident increases 15 percent. But in companies you get the opposite effect. As companies get bigger, they usually get more bureaucratic and less innovative per employee.

The mayor of a city doesn’t tell its residents what to do or where to live; there is a certain infrastructure that a city must provide, such as the grid: water, power, and sewage. And there are certain basic laws that a city enforces. But for the most part, what happens when a city grows and innovates is a result of the self-organization that happens with a city’s residents, businesses, and other organizations.

The Quarterly: What are the key principles of Zappos?

Tony Hsieh: The way I think about it, there are three different pillars, or dimensions, that are foundational to us at Zappos. And we need to make sure that all are working well.

The first is culture and values. Now, we’re not saying that other companies should adopt our values. One of the interesting things I’ve learned from the research is that it actually doesn’t matter what your values are. What matters is that you have them and you commit to them and align the entire organization around them. That means you’re willing to hire and fire based on them. Most large companies have things called core values or guiding principles and so on. But I think what employees find at those companies is that the principles read like PR statements. You see them on the company’s website, and maybe on the lobby wall in reception, but then no one really pays attention to them.

We have ten core values that serve as a formalized definition of our culture. And these ten were crowdsourced—I asked our employees what our values should be. Then we went back and forth for about a year and came up with our ten. They’ve become part of our culture. They’re part of our employees’ everyday language. Our core values actually just come up naturally in everyday conversation. And once they become part of the language, they become part of the mind-set. So that’s our first pillar: values alignment.

The second pillar is purpose. Both on the individual level and on the organizational level, we are very explicit about purpose statements. And one of the things that holacracy enables is a hierarchy of purpose statements. There’s the purpose of the company, which for us is “To Live and Deliver WOW.” We have something we refer to as the general company circle [GCC]that holds the company purpose statement. And within that circle, there are subcircles and roles, and we cascade down from there into a hierarchy of more subcircles and roles. But you can pick any role anywhere in the hierarchy, and there’s an entire set of purpose statements that all link ultimately back up to the company purpose. The purpose statements are something that we do think about occasionally, but I’d like us to get to the point where our purpose hierarchy is as top of mind as our core values. Right now, purpose statements aren’t part of our everyday language, at least the way our core values are. I think that’s a big opportunity for us.

The third pillar I’ll call “market-based dynamics.” Just like in a city, it’s important to have a true market, to break up monopolies, and to have different internal teams become customers of each other. We’re building an internal currency as well as the internal tools and systems to support an underlying infrastructure to allow for multiple participants, fast feedback loops, and things like crowdsourced participation. Imagine the equivalent of, say, the stock market, but for inventory purchasing. What if employees or teams could bet on inventory the same way people and organizations bet on the stock market in the real world today?

Video
The Zappos ecosystem

The Quarterly: So you’re trying to unleash the power of the organization as a market-based system, and also the potential of each person who makes up that market.

Tony Hsieh: Yes. One of the learnings we’ve had about self-organization and self-management is that it’s not just a systems change; it’s also a personal journey for each individual employee. Self-organization and self-management is about the entrepreneurial mind-set. A study was done several years ago that looked at what separated the great entrepreneurs from the mediocre ones. They found that the great entrepreneurs highly overindex for three characteristics: first, being comfortable with ambiguity; second, having a strong sense of curiosity; and third—not as high, but still overindexed—was emotional intelligence. And I think under self-organization, these three characteristics are even more important for us to hire for at Zappos.

A market is able to incorporate the intelligence of all the individual players and get feedback much faster than most other feedback mechanisms. I don’t watch a lot of sports—except the Super Bowl—but even if I don’t know who’s playing, I do know that the sports-betting market, more than any other method that I’m aware of, does an amazing job of telling you what the right odds are. It does that by using the collective intelligence of the group.

To harness collective intelligence, we think of every single employee as a human sensor. Everyone senses different things, and you want a way to process all of that input. An airplane is one analogy. There are all of these different sensors. Some sensors, like the altimeter, are probably more important than others, but you want to be aware of all of them. Even if the altimeter looks fine, and most of the other sensors look fine, that doesn’t mean it’s OK to ignore the low-voltage warning light when it turns on. You don’t allow the other sensors to outvote the low-voltage warning light and ignore it, yet the analogous thing happens all the time in organizations.

Some intern might say, “Hey, there’s this Instagram thing we should pay attention to,” but no one senior has heard of it, so the suggestion gets ignored. The structure that we have enables that intern—and basically all of our employees, if you think of them as human sensors—to actually do something about it, versus just getting outvoted. The idea is that everyone is a human sensor capable of sensing tensions, and it’s important to note that a tension is not necessarily a bad thing. It’s just the gap between the way things are and what’s possible. Zappos is currently at 1,500 employees; everyone is able to actually sense tensions, and collectively our system is designed to process them. When you do that, you’ll notice new tensions, and if everyone consistently does that, that’s going to move the organization forward.

The Quarterly: How does this work in practice?

Tony Hsieh: In our organization, your tensions are based on the purpose statements for the roles that you hold or the circles you’re in. And employees can belong to multiple circles. Our 1,500 employees right now are in about 500 circles, and if you want to be kept updated on what’s going on in the circle meetings, you can subscribe to any of them, all of them, or just the ten or so circles that you’re interested in. For example, let’s say there’s a circle that was originally focused on our campus, and the original purpose statement of the campus team was to have an amazing experience for employees. Then, someone else can bring up, “Well, as we’ve evolved over the years, we really want to be inclusive of the surrounding community as well.” They can modify the purpose of the campus circle so that it’s an amazing experience for employees, vendors, or someone else in our community, and that experience will help elevate the Zappos brand. And then, someone else can say, “Well, what about this other thing that you’ve forgotten?” So that can get added to the purpose statement or subtracted from it over time.

There are also elements that are similar to the Hollywood model. Certain people get more inspired working around other people and just kind of play off of each other’s ideas. They work on projects, and then people who enjoyed working with each other decide to work together on the next project. And then, over time, what evolves is a dream team that works on the best movies.

The Quarterly: It sounds like a key element is tapping into the dynamics of how people work together.

Tony Hsieh: One term we use is “collisions.” It’s about how often you run into different, diverse people that you might have a random conversation with and ultimately end up collaborating with. We think about it in terms of density, as well. The average density of office space in the US is about 300 square feet per employee—including hallways, conference rooms, et cetera. We’re at about 100 square feet per employee here at Zappos, and the reason is because research has shown that if you sit twice as far away from someone in an office environment, you don’t see them half as often—you see them half as often squared, so a quarter as often.

It’s also important to understand that the best-performing teams are not created by simply putting together the best-performing individuals. Have you heard of the “super chicken” research? With chickens, you can measure productivity through how many eggs they lay. And so, in this study, Strategy One was to breed, say, ten chickens in a cage, find the best-producing ones, breed them for the next generation, and then see what happens six or seven generations down the line. And what they found was that at the end of the six or seven generations, Strategy One had these super, alpha chickens, and any one of them was an amazing producer. The problem is they killed half of the other chickens in the cage. And so, as a cage, they didn’t produce as much as Strategy Two, which was to breed for the best-performing cage. Within that best-performing cage, there might be a chicken that maybe doesn’t lay that many eggs but is the one that keeps the peace amongst all the other chickens. And if you’re trying to maximize for overall productivity, you want to go for Strategy Two. But most big corporations go for Strategy One.

The Quarterly: In terms of corporate structure, what do you think you give up by optimizing for adaptability instead of predictability?

Tony Hsieh: There’s a quote attributed to Charles Darwin—it may be misattributed—but it’s something like, “It’s not the fastest or strongest or most intelligent of species that survives. It’s the one most adaptable to change.” The world’s moving faster and faster. Technology is enabling things to happen more and more quickly, and information flows much more quickly than it did 20 to 30 years ago. It’s really going from a mind-set of, “How do we try to predict, plan, and control and execute on a specific plan?” to a mind-set that’s more about, “How can we get fast feedback loops? How do we constantly sense and respond and build the organization around adaptability and resilience and longevity?” versus the more traditional mind-set of efficiency. Systems theory and research has shown that if you’re maximizing for both efficiency and stability, usually that’s at the cost of resilience. Resilience is harder to maximize for because it’s harder to observe and measure compared to efficiency and/or stability.

It may be that, on a metalevel, what’s less predictable is what our org chart is going to look like six months from now. But I don’t know whether predictability is actually an advantage. I think that’s how organizations get stuck, because they want that predictability of structure. But if it’s the wrong structure, what’s the benefit of being predictably wrong? The structure of the organization is a variable that affects the productivity and output of the organization. And most organizations aren’t designed for changing their org structure in any efficient or easy way.

The Quarterly: Speaking of variables, your core business is very data- and analytics-enabled. How do data and analytics play into your organizational thinking? Do you use them to assess performance?

Tony Hsieh: It depends. We have a team that looks at data quite a bit. There are certain things where we know that if we spend, say, X dollars in paid advertising in this channel, we’re going to get Y dollars in sales. So we should do more of that, as long as the ROI continues to make sense.

But you can’t put everything in a metric. There are certain people that I get more inspired working with, where we just kind of play off of each other’s ideas. I wouldn’t even know how to put that into some sort of metric form that passes legal and HR.

We believe that employees are much more than just what their specific job description is. Maybe it’s through volunteering at an event, or at the company all-hands meetings where an employee that’s great at dancing can go do that even though that has nothing to do with their job description. But beyond just the hobby aspect, I think there’s so much creative potential and intelligence that each individual employee has. We’re trying to figure out how to create the best structure that releases as much of that as possible. Most structures just end up constraining, so you end up getting 10 percent of a person’s potential versus, hopefully, close to 100 percent.

More than ten years ago, we made a commitment to our core values. Committing means we’re actually willing to hire and fire people based on whether they’re living the Zappos core values, completely independent of their actual job performance. Our hiring team interviews for the normal stuff, but then our HR recruiting team does a separate set of interviews purely for culture and values fit. Candidates have to pass both in order to be hired. We’ve passed on a lot of really smart, talented people that we know, in the short term, can make an immediate impact on our top or bottom line. But if they’re not a fit for our values, then we won’t hire them—and for that reason alone. It’s the same thing for firing. If an employee is not living up to our values, even if they’re the top salesperson or whatever, most companies would say, “Well, this guy is kind of a jerk, but, you know, we’ll let it slide because he’s bringing in the revenue.” Whereas for us, it wouldn’t even be a discussion.

The Quarterly: You seem very comfortable with the fact that you don’t quite know where this ship is headed, that it’s steering itself.

Tony Hsieh: We’ve all been brought up to believe that “If I do X, then Y is going to happen.” But there are always unintended consequences. If you look at the great inventions, they emerge more from nonlinear ways of thinking. It’s not, “Oh, I need to go invent X, and so all I have to do is steps one through ten, and then, all of a sudden, we come up with this random invention.” I think people fool themselves into thinking things can be predicted and, therefore, controlled.

And so at Zappos, the bar is: Is it safe enough to try? It doesn’t matter if other employees think it’s a bad idea. I can take that input. But is it safe enough to try? At most companies, including us, historically, it ended up being more about consensus building—which is great when you’re small. But consensus building doesn’t scale. Self-organization, if done right, does scale.

About the author(s)

Tony Hsieh is the CEO of Zappos. This interview was conducted by Aaron De Smet, a senior partner in McKinsey’s Houston office, and Chris Gagnon, a senior partner in the New Jersey office.
More from McKinsey Quarterly
Report

Remaking the bank for an ecosystem world

Interactive - McKinsey Quarterly

Five Fifty: The digital effect

Commentary - McKinsey Quarterly

Wellness at work: The promise and pitfalls