by Jake Knapp
We’d love to tell you that we, the authors, were the genius heroes of this story. It’d be wonderful if we could swoop into any company and dish out brilliant ideas that would transform it into a breakout success. Unfortunately, we are not geniuses. Savioke’s sprint worked because of the real experts: the people who were on the team all along. We just gave them a process to get it done.
Here’s how the Savioke sprint went down. And if you’re not a roboticist yourself, don’t worry. We use this same exact sprint structure for software, services, marketing, and other fields.
First, the team cleared a full week on their calendars. From Monday to Friday, they canceled all meetings, set the “out of office” responders on their email, and completely focused on one question: How should their robot behave around humans?
Next, they manufactured a deadline. Savioke made arrangements with the hotel to run a live test on the Friday of their sprint week. Now the pressure was on. There were only four days to design and prototype a working solution.
On Monday, Savioke reviewed everything they knew about the problem. Steve talked about the importance of guest satisfaction, which hotels measure and track religiously. If the Relay robot boosted satisfaction numbers during the pilot program, hotels would order more robots. But if that number stayed flat, or fell, and the orders didn’t come in, their fledgling business would be in a precarious position.
Together, we created a map to identify the biggest risks. Think of this map as a story: guest meets robot, robot gives guest toothbrush, guest falls for robot. Along the way were critical moments when robot and guest might interact for the first time: in the lobby, in the elevator, in the hallway, and so on. So where should we spend our effort? With only five days in the sprint, you have to focus on a specific target. Steve chose the moment of delivery. Get it right, and the guest is delighted. Get it wrong, and the front desk might spend all day answering questions from confused travelers.
One big concern came up again and again: The team worried about making the robot appear too smart. “We’re all spoiled by C-3PO and WALL-E,” explained Steve. “We expect robots to have feelings and plans, hopes and dreams. Our robot is just not that sophisticated. If guests talk to it, it’s not going to talk back. And if we disappoint people, we’re sunk.”
On Tuesday, the team switched from problem to solutions. Instead of a raucous brainstorm, people sketched solutions on their own. And it wasn’t just the designers. Tessa Lau, the chief robot engineer, sketched. So did Izumi Yaskawa, the head of business development, and Steve, the CEO.
By Wednesday morning, sketches and notes plastered the walls of the conference room. Some of the ideas were new, but some were old ideas that had once been discarded or never thought through. In all, we had twenty-three competing solutions.
How could we narrow them down? In most organizations, it would take weeks of meetings and endless emails to decide. But we had a single day. Friday’s test was looming, and everybody could sense it. We used voting and structured discussion to decide quickly, quietly, and without argument.
The test would include a slate of Savioke designer Adrian Canoso’s boldest ideas: a face for the robot and a soundtrack of beeps and chimes. It would also include one of the more intriguing but controversial ideas from the sketches: When the robot was happy, it would do a dance. “I’m still nervous about giving it too much personality,” Steve said. “But this is the time to take risks.”
“After all,” said Tessa, “if it blows up now, we can always dial back.” Then she saw the looks on our faces. “Figure of speech. Don’t worry, the robot can’t actually blow up.”
As Thursday dawned, we had just eight hours to get the prototype ready for Friday’s live test in the hotel. That shouldn’t have been enough time. We used two tricks to finish our prototype on time:
1. Much of the hard work had been done already. On Wednesday, we had agreed on which ideas to test, and documented each potential solution in detail. Only the execution remained.
2. The robot didn’t need to run autonomously, as it would eventually in the hotel. It just needed to appear to work in one narrow task: delivering one toothbrush to one room.
Tessa and fellow engineer Allison Tse programmed and tuned the robot’s movements using a beat-up laptop and a PlayStation controller. Adrian put on a pair of massive headphones and orchestrated the sound effects. The “face” was mocked up on an iPad and mounted to the robot. By 5 p.m., the robot was ready.
For Friday’s test, Savioke had lined up interviews with guests at the local Starwood hotel in Cupertino, California. At 7 a.m. that morning, we rigged a makeshift research lab inside one of the hotel’s rooms by duct-taping a couple of webcams to the wall. And at 9:14 a.m., the first guest was beginning her interview.
• • •
The young woman studied the hotel room decor: light wood, neutral tones, a newish television. Nice and modern, but nothing unusual. So what was this interview all about?
Standing beside her was Michael Margolis, a research partner at GV. For now, Michael wanted to keep the subject of the test a surprise. He had planned out the entire interview to answer certain questions for the Savioke team. Right now, he was trying to understand the woman’s travel habits, while encouraging her to react honestly when the robot appeared.
Michael adjusted his glasses and asked a series of questions about her hotel routine. Where does she place her suitcase? When does she open it? And what would she do if she’d forgotten her toothbrush?
“I don’t know. Call the front desk, I suppose?”
Michael jotted notes on a clipboard. “Okay.” He pointed to the desk phone. “Go ahead and call.” She dialed. “No problem,” the receptionist said. “I’ll send up a toothbrush right away.”
As soon as the woman returned the receiver to its cradle, Michael continued his questions. Did she always use the same suitcase? When was the last time she’d forgotten something on a trip?
Brrrring. The desk phone interrupted her. She picked up, and an automated message played: “Your toothbrush has arrived.”
Without thinking, the woman crossed the room, turned the handle, and opened the door. Back at headquarters, the sprint team members were gathered around a set of video displays, watching her reaction.
“Oh my god,” she said. “It’s a robot!”
The glossy hatch opened slowly. Inside was the toothbrush. The robot made a series of chimes and beeps as the woman confirmed delivery on its touch screen. When she gave the experience a five-star review, the little machine danced for joy by twisting back and forth.
“This is so cool,” she said. “If they start using this robot, I’ll stay here every time.” But it wasn’t what she said. It was the smile of delight that we saw over the video stream. And it was what she didn’t do—no awkward pauses and no frustration as she dealt with the robot.
Watching the live video, we were nervous throughout that first interview. By the second and third, we were laughing and even cheering. Guest after guest responded the same way. They were enthusiastic when they first saw the robot. They had no trouble receiving their toothbrushes, confirming delivery on the touch screen, and sending the robot on its way. People wanted to call the robot back to make a second delivery, just so they could see it again. They even took selfies with the robot. But no one, not one person, tried to engage the robot in any conversation.
At the end of the day, green check marks filled our whiteboard. The risky robot personality—those blinking eyes, sound effects, and, yeah, even the “happy dance”—was a complete success. Prior to the sprint, Savioke had been nervous about overpromising the robot’s capability. Now they realized that giving the robot a winsome character might be the secret to boosting guest satisfaction.
Not every detail was perfect, of course. The touch screen was sluggish. The timing was off on some of the sound effects. One idea, to include games on the robot’s touch screen, didn’t appeal to guests at all. These flaws meant reprioritizing some
engineering work, but there was still time.
Savioke’s Relay robot.
Three weeks later, the robot went into full-time service at the hotel. And the Relay was a hit. Stories about the charming robot appeared in the New York Times and the Washington Post, and Savioke racked up more than 1 billion media impressions in the first month. But, most important, guests loved it. By the end of the summer, Savioke had so many orders for new robots that they could hardly keep up with production.
Savioke gambled by giving their robot a personality. But they were only confident in that gamble because the sprint let them test risky ideas quickly.
The trouble with good ideas
Good ideas are hard to find. And even the best ideas face an uncertain path to real-world success. That’s true whether you’re running a startup, teaching a class, or working inside a large organization.
Execution can be difficult. What’s the most important place to focus your effort, and how do you start? What will your idea look like in real life? Should you assign one smart person to figure it out or have the whole team brainstorm? And how do you know when you’ve got the right solution? How many meetings and discussions does it take before you can be sure? And, once it’s done, will anybody care?
As partners at GV, it’s our mission to help our startups answer these giant questions. We’re not consultants paid by the hour. We’re investors, and we succeed when our companies succeed. To help them solve problems quickly and be self-sufficient, we’ve optimized our sprint process to deliver the best results in the least time. Best of all, the process relies on the people, knowledge, and tools that every team already has.
Working together with our startups in a sprint, we shortcut the endless-debate cycle and compress months of time into a single week. Instead of waiting to launch a minimal product to understand if an idea is any good, our companies get clear data from a realistic prototype.
The sprint gives our startups a superpower: They can fast-forward into the future to see their finished product and customer reactions, before making any expensive commitments. When a risky idea succeeds in a sprint, the payoff is fantastic. But it’s the failures that, while painful, provide the greatest return on investment. Identifying critical flaws after just five days of work is the height of efficiency. It’s learning the hard way, without the “hard way.”
At GV, we’ve run sprints with companies like Foundation Medicine (makers of advanced cancer diagnostics), Nest (makers of smart home appliances), and Blue Bottle Coffee (makers of, well, coffee). We’ve used sprints to assess the viability of new businesses, to make the first version of new mobile apps, to improve products with millions of users, to define marketing strategies, and to design reports for medical tests. Sprints have been run by investment bankers looking for their next strategy, by the team at Google building the self-driving car, and by high school students working on a big math assignment.
This book is a DIY guide for running your own sprint to answer your pressing business questions. On Monday, you’ll map out the problem and pick an important place to focus. On Tuesday, you’ll sketch competing solutions on paper. On Wednesday, you’ll make difficult decisions and turn your ideas into a testable hypothesis. On Thursday, you’ll hammer out a realistic prototype. And on Friday, you’ll test it with real live humans.
Instead of giving high-level advice, we dig into the details. We’ll help you assemble the perfect sprint team from the people with whom you already work. You’ll learn big stuff (like how to get the most out of your team’s diverse opinions and one leader’s vision), medium stuff (like why your team should spend three straight days with your phones and computers off), and nitty-gritty stuff (like why you should eat lunch at 1 p.m.). You won’t finish with a complete, detailed, ready-to-ship product. But you will make rapid progress, and know for sure if you’re headed in the right direction.
You’ll see some methods that look familiar and others that are new. If you’re familiar with lean development or design thinking, you’ll find the sprint is a practical way to apply those philosophies. If your team uses “agile” processes, you’ll find that our definition of “sprint” is different, but complementary. And if you haven’t heard of any of these methods, don’t worry—you’ll be fine. This is a book for experts and beginners alike, for anyone who has a big opportunity, problem, or idea and needs to get started. Every step has been tried, tweaked, tested, and measured over the course of our 100+ sprints and refined with the input we’ve gathered from the growing sprint community. If it doesn’t work, it’s not in the book.
At the end, you’ll find a set of checklists, including a shopping list and day-by-day guides. You don’t have to memorize everything now—the checklists await you once you’re ready to run your own sprint. But before you start that sprint, you’ll need to plan carefully to make it a success. In the next chapters, we’ll show you how to set the stage.
Set the Stage
Before the sprint begins, you’ll need to have the right challenge and the right team. You’ll also need time and space to conduct your sprint. In the next three chapters, we’ll show you how to get ready.
1
Challenge
In 2002, a clarinet player named James Freeman quit his job as a professional musician and founded . . . a coffee cart.
James was obsessed with freshly roasted coffee. In those days in the San Francisco area, it was nearly impossible to find coffee beans with a roast date printed on the bag. So James decided to do it himself. He carefully roasted beans in a potting shed at home, then drove to farmers’ markets in Berkeley and Oakland, California, where he brewed and sold coffee by the cup. His manner was polite and accommodating, and the coffee was delicious.
Soon James and his cart, called Blue Bottle Coffee, developed a following. In 2005, he established a permanent Blue Bottle location in a friend’s San Francisco garage. Over the next few years, as the business grew, he slowly opened more cafés. By 2012, Blue Bottle had locations in San Francisco, Oakland, Manhattan, and Brooklyn. It was a business that many would have considered perfect. The coffee was ranked among the best nationwide. The baristas were friendly and knowledgeable. Even the interior design of the cafés was perfect: wooden shelves, tasteful ceramic tiles, and an understated logo in the perfect shade of sky blue.
But James didn’t consider the business perfect, or complete. He was still just as passionate about coffee and hospitality, and he wanted to bring the Blue Bottle experience to even more coffee lovers. He wanted to open more cafés. He wanted to deliver freshly roasted coffee to people’s homes, even if they didn’t live anywhere near a Blue Bottle location. If that coffee cart had been Sputnik, the next phase would be more like a moon shot.
So in October 2012, Blue Bottle Coffee raised $20 million from a group of Silicon Valley investors, including GV. James had many plans for that money, but one of the most obvious was building a better online store for selling fresh coffee beans. But Blue Bottle wasn’t a tech company and James was no expert at online retail. How could he translate the magic of his cafés to smartphones and laptops?
Several weeks later, on a bright December afternoon, Braden Kowitz and John Zeratsky met up with James. They sat around a counter, drank coffee, and discussed the challenge. The online store was important to the company. It would take time and money to get it right, and it was difficult to know where to start. In other words, it sounded like a perfect candidate for a sprint. James agreed.
They talked about who should be in the sprint. An obvious choice was the programmer who would be responsible for building Blue Bottle’s online store. But James also included Blue Bottle’s chief operating officer, chief finance officer, and communications manager. He included the customer service lead who handled questions and complaints. He even included the company’s executive chairman: Bryan Meehan, a retail expert who started a chain of organic grocery stores in the UK. And, of course, James himself would be in the room.
The online store was essentially a s
oftware project—something our team at GV was very familiar with. But this group looked almost nothing like a traditional software team. These were busy people, who would be missing a full week of important work. Would the sprint be worth their time?
• • •
On Monday morning of our sprint week, the Blue Bottle team gathered in a conference room at GV’s office in San Francisco. We made a diagram on the whiteboard showing how coffee buyers might move through the online store. The Blue Bottle team targeted a new customer purchasing coffee beans. James wanted to focus the sprint on this scenario because it was so difficult. If they could establish credibility and create a great experience for someone who had never heard of Blue Bottle, let alone visited their cafés or tasted their coffee, then every other situation should be easy by comparison.
We ran into a big question: How should we organize the coffee? The shopper in this scenario would be choosing between a dozen or so varieties of bean, each in a nearly identical bag. And—unlike in Blue Bottle’s cafés—there would be no barista there to help choose.
At first, the answer seemed obvious. From boutique coffee roasters to mainstream giants like Starbucks, retailers tend to organize coffee by the geographic region where it was grown. Africa, Latin America, the Pacific. Honduran coffee vs. Ethiopian coffee. It would be logical for Blue Bottle to categorize their beans the same way.
“I have to admit something,” Braden announced. Everyone turned. “I’m into coffee, okay? I have a scale at home and everything.” Electronic scales are the hallmark of a true coffee freak. Owning a scale meant Braden weighed the water and coffee beans so that he could experiment and adjust ratios as he brewed. We’re talking science here. Coffee scales are accurate to a fraction of a gram.