The Startup Way: How Modern Companies Use Entrepreneurial Management to Transform Culture and Drive Long-Term Growth

by Eric Ries

the goal is to make it possible for startup teams to operate reliably and give every employee the opportunity to act in an entrepreneurial way.

The ultimate goal of the Startup Way is for organizations to be in a state of continuous transformation, which will allow them to flourish in any circumstance.

The more corporate innovators I met, the more I heard about how much faith their bosses put in forecasts as a tool for holding people accountable—even senior managers who (I thought) surely would know better. The “fantasy plan” of the original pitch is often far too optimistic to be used as a real forecast. But managers, lacking any other system to use, need something to hold on to. Without an alternative, they cling to the forecast—even if it’s just made up. You’ve probably started to sense the problem here: An older system of accountability, designed in a very different time and for a very different context, is still being used in situations where it doesn’t work. Sometimes, failure to hit the forecast means a team executed poorly. But sometimes it means the forecast itself was a fantasy. How can we tell the difference?

One day, as I was meeting with a Six Sigma black belt from one of GE’s industrial businesses—who was quite skeptical— I found myself distracted by the mug on his desk, which read: FAILURE IS NOT AN OPTION. Nobody in the startup world could have such a mug, I mused; it would be ridiculous. My experience is full of situations where reality proved too unpredictable to avoid failure. I thought of the best, most successful entrepreneurs I know. What would their mug say? I settled on: I EAT FAILURE FOR BREAKFAST. The tension between those two slogans is a great starting point for understanding not only why startups have had such a hard time adopting traditional management methods, and vice versa—but also what connects them. There was a time when producing high-quality products on time, on budget, and at scale was one of the preeminent problems of the age. Understanding how to build quality into products from the inside out required mastering the new statistical science of variation, and then devising tools, methodologies, and training programs that could make doing so practical. Standardization, mass production, lean manufacturing, and Six Sigma are all fruits of this hard-won conceptual victory.

As Jeff Bezos said at the time: I’ve made billions of dollars of failures at Amazon.com. Literally. None of these things are fun, but they also don’t matter. What matters is that companies that don’t continue to experiment or embrace failure eventually get in the position where the only thing they can do is make a Hail Mary bet at the end of their corporate existence. I don’t believe in bet-the-company bets.

One of my favorite stories about Sheryl Sandberg, Facebook’s dynamic COO, comes from a company meeting in which employees were complaining about the “unfairness” of having their performance evaluated based on the success of the projects they had worked on, rather than just their individual contribution to those projects. Sandberg acknowledged these concerns, but her reply has stuck with me for years. She asked the employees to imagine a favorite company of theirs that had been disrupted. Kodak, say, or RIM. Imagine all the employees of that doomed company, in the months and years leading up to the disruption. Think of all the people who got positive reviews, got promoted, got paid bonuses for their excellent functional contribution—all while the company went down to ignoble defeat. Do you really want to be one of those managers?

A modern company is one that has both halves, both systems. It has a capacity to produce products with great reliability and quality, but also to discover what new products to produce.

AN OLD-FASHIONED COMPANY is founded on steady growth through prescriptive management and controls, and is subject to tremendous pressure to perform in short-term intervals such as quarterly reports. A MODERN COMPANY is founded on sustained impact via continuous innovation, and focused on long-term results. AN OLD-FASHIONED COMPANY is made up of experts in specialized functional silos, between which work passes in a stage-gate or waterfall process that sends projects from function to function with specific milestones tied to each handover. A MODERN COMPANY is made up of cross-functional teams that work together to serve customers through highly iterative and scientific processes. AN OLD-FASHIONED COMPANY tends to operate huge programs. A MODERN COMPANY operates rapid experiments. AN OLD-FASHIONED COMPANY uses internal functions such as legal, IT, and finance to mitigate risk through compliance with detailed procedures. A MODERN COMPANY uses internal functions to help its employees meet their goal of serving customers, sharing the responsibility to drive business results. AN OLD-FASHIONED COMPANY prioritizes even highly uncertain projects based on ROI, traditional accounting, and market share. To measure success, project teams track and share numbers designed to look as good as possible (“vanity metrics”)—but not necessarily to reveal the truth. A MODERN COMPANY attempts to maximize the probability and scale of future impact. Project teams report and measure leading indicato...

That’s why, in the tech industry especially, small teams put a huge premium on reusing existing technology and assembling products out of preexisting components. More than at any other time in history, these components can be combined without requiring explicit permission or a business-development relationship.

can’t tell you how many times I’ve worked with teams in traditional enterprises that literally don’t know what problem they are trying to solve from the customer’s point of view.

For many internal projects—in IT, HR, and finance—as well as products sold via third-party distributors, people often don’t know what the word customer even means. One IT team in a large enterprise I worked with refused to see the employees who used their product as customers. They insisted that employees had no choice and that IT could mandate the usage of any product. But the word customer always implies a choice in the matter, so rather than argue, we decided to go see for ourselves. I asked the team to interview several existing users of the IT system within the firm to see what their rate of compliance was. The team was scandalized to learn that employees hated the software so much that they were using a wide array of work-arounds to avoid using it at all. Some employees were actually duplicating the system’s calculations by hand on paper. Customers, even internal ones, always have a choice. No corporate mandate can ever hope to achieve 100 percent compliance unless employees buy in.

This was my chance to explain how we do risk mitigation in Silicon Valley, using something called metered funding. (This is the opposite of the typical corporate budgeting approach, which I call entitlement funding and discuss in Chapter 7.)

As Mark Zuckerberg says in his famous manifesto (in Facebook’s S-1 filing): “Try to build the best services over the long term by quickly releasing and learning from smaller iterations rather than [by] trying to get everything right all at once….We have the words Done is better than perfect painted on our walls to remind ourselves to always keep shipping.”5

As it turned out, there was absolutely no physical, organizational, political, or regulatory barrier to doing an experiment that very day. The only impediment was the ingrained force of habit that led this team to overlook this simple solution. As you’ll see in the next few chapters, this is incredibly common. The problem isn’t that the engineers can’t figure out how to build an MVP; it’s that they’ve never considered it worth doing. When the mental landscape is shaped by management habits, it’s hard to have a breakthrough.

My suggestion is: Schedule the pivot-or-persevere meeting in advance. Put it on the calendar. Make it a routine part of everyday life. Find the right cadence for the startup in question. I usually recommend about once every six weeks—certainly not more often than once a month or less frequently than once a quarter. Have the meeting, and let it be no big deal. It’s not an existential crisis. It’s not an admission that we don’t know what we are doing. It’s just a chance to ask ourselves: What evidence do we have that our current strategy is taking us closer to our vision?

It turns out that when the product managers told me the customers wanted more buttons, they weren’t actually talking about the consumer, because the consumer is not their customer. “The customer is the category buyer at the big-box retail stores where we sell most of our appliances,” they explained. “Our sales team spends extensive time with the customer to find out what they want us to build.”

By the time a company is into Phase Two of transformation, it will have two major sources of information to draw upon. This is the payoff for letting teams fail and succeed. The first source consists of all the exceptions that had to be made for a team to kick off its project: compliance, hiring, approvals, etc., many of which we touched upon in Chapter 6. What were the biggest issues the teams faced? The variety of problems tends to be far-reaching, but the message among them is clear: What was done ad hoc and by decree in the early stages must now be systematized. The other source is the results of the early projects themselves. Some will have succeeded and become templates for others to emulate. Many will have failed and produced detailed information as to why.

Because sharing information about new methods throughout the organization is a key part of the second phase (as we did at GE through the trainings, materials developed, and, ultimately, the GE Beliefs), the USDS created the Digital Services Playbook, which the team released publicly on the USDS website the day the organization was officially launched. A list of thirteen key “plays” taken from the private sector and government alike, the playbook offers yet another way for the organization to fan its methods out into government.2

Considering the ways in which his company has changed, Jeff Immelt recently said to me: “One of the things that makes FastWorks a little bit different than other things we’ve done is that certain functions in the company could stop it. You could say, if you’re one of the enabling functions, ‘We don’t have the budget,’ or, ‘I worry about compliance.’ ” The solution? “You need messaging, not so much for the people who are the practitioners, but for the people who can stop things. You put them on notice and say, ‘You’re going to fail. You do that at your own peril.’ I think that’s where culture change is hard, because you’ve got a movement, and the movement can go quickly, but you’re not just trying to get people to come with you. You’ve got to stop the people who want to block it.”

The team was ready to explore their ideas, but one of the challenges they faced was that they were addressing two different customers, one of which was internal: Intuit’s phone support staff, who help the company’s external customers with billing issues. The team decided to run, with the internal phone support staff, small experiments related to everything from tone of voice, to billing communication, to more traditional features. The end result was that the team made rapid progress and reframed their understanding of this staff as “potential startup customers,” in Blank’s words. This small change in framing had a big effect on the team’s approach as they began to apply Lean Startup principles to this new “customer.”

Before the company instituted lean and agile methods, there were thirteen layers between CIO Jeff Smith and the first-line leader in the squads (what IBM calls teams). Now there are five. Smith says the coaches are the accelerators.

Then, referring to some words of wisdom given to him by Elon Musk, the founder of Tesla, who had described for Chesky the three “eras” of a startup—creation, building, and administration—Chesky added: “Airbnb will never be in the administration era. It will always be in a building era.”2

They remained focused on co-located, cross-functional teams and “short sprints of exposing functionality to the users,” he says. “We would take off-the-shelf functionality and go directly into testing mode.” This was a complete reversal from the old method. Now the onus was on the customers—GE’s internal customers—to show the team why something didn’t work well for them. “That was sort of the key to it,” says Richards. “We’d assume that the enterprise software we’re deploying works…then we will only fix or configure those things that caused us a compliance problem or caused massive inefficiencies in the user base.” Once they were nearly finished with one module, the team would move on to the next one. Working in this way, they reduced their deployment time from eighteen months to between four and six months.

Innovation Accounting (IA) is a way of evaluating progress when all the metrics typically used in an established company (revenue, customers, ROI, market share) are effectively zero. It provides a framework of chained leading indicators, each of which predicts success. Each link in the chain is essential and, when broken, demands immediate attention. It’s a focusing device for teams, keeping their attention on the most important leap-of-faith assumptions. It’s a common, mathematical vocabulary for negotiating the use of resources among competing functions, divisions, or regions. It provides a way to tie long-term growth and R&D into a system that follows a clear process for funding innovation that can be audited for its ability to drive value creation.

What percentage of projects are canceled, and how long does it take to stop them? Before growth boards: Only 10 percent of the division’s projects were being killed. That meant 90 percent of the projects delivered something, regardless of whether or not someone wanted it. First round of growth boards: 20 percent of the projects were eliminated after a ninety-day cycle and for a lot less money. Second round of growth boards: 50 percent of the projects were stopped, many of them after only a sixty-day cycle. How are projects being killed? Before growth boards: They weren’t, for the most part, being canceled, for all the reasons we’ve discussed thus far. First round of growth boards: Projects were terminated by the growth board. Second round of growth boards: The responsibility shifted to the teams. “The teams would come in, and they would almost pitch the case like they wanted us to stop them.” Third round of growth boards (after the seed money had been handed down to product companies): The teams would come to the board and say, “We’ve already stopped the project.”

Yet this is what too many of our public companies are doing: cannibalizing their long-term value by destroying their own brand, squeezing vendors, shortchanging customers, failing to invest in employees, and using the company’s resources to enrich insiders and activist investors via financial engineering. All of these activities share the same problem: They work only in the short term. In companies that have grown a sufficiently large and productive “forest” over years or decades, there’s an awful lot of firewood to be cut down before the damage becomes evident. This is the inevitable result of treating companies as if their obligation to maximize shareholder value means maximizing quarterly returns.