Inspired by the conversations with Hart and others about voice computing, he emailed Hart, device vice president Ian Freed, and senior vice president Steve Kessel on January 4, 2011, linking the two topics: “We should build a $20 device with its brains in the cloud that’s completely controlled by your voice.” It was another idea from the boss who seemed to have a limitless wellspring of them.

Characteristically secretive, Amazon has never revealed the name of the voice artist behind Alexa. I learned her identity after canvasing the professional voice-over community: Boulder-based singer and voice actress Nina Rolle. Her professional website contained links to old radio ads for products such as Mott’s Apple Juice and the Volkswagen Passat—and the warm timbre of Alexa’s voice was unmistakable. Rolle said she wasn’t allowed to talk to me when I reached her on the phone in February of 2021.

Another early Doppler employee later speculated that Bezos’s famous lack of sophisticated musical tastes played a role in his reaction. When Bezos was testing an early Doppler unit, for example, he asked it to play one of his favorite songs: the theme to the classic TV show Battlestar Galactica. “Jeff was pushing really hard to make sure this product was more than just music,” said Ian Freed, Greg Hart’s boss. “He wouldn’t let go of it being a more generalized computer.”

Bezos also had some worrisome blind spots about smartphones. “Does anyone actually use the calendar on their phone?” he asked in one meeting. “We do use the calendar, yes,” someone who did not have several personal assistants replied.

Working eighty to ninety hours a week, [Echo team] employees were missing whole chunks of their family’s lives, and Bezos wasn’t letting up. He wanted to see everything and made impetuous new demands. On an unusually clear Seattle day, with the setting sun streaming through his conference room window, for example, Bezos noticed that the ring’s light was not popping brightly enough, so he ordered a complete redo.

But there were drawbacks to the frenetic speed and growth. For years the Alexa smartphone app looked like something a design student had come up with during a late-night bender.

One of [Gianna] Puerini’s first tasks was selecting a code name so boring that no one would pay any attention to them. For the next few years, the [Amazon Go grocery store] team would go by the initials, IHM, short for the nonsensical “Inventory Health Management.”

They were entering the hottest crucible at Amazon—a Jeff project, one that had the attention of the ever-curious and demanding boss. During a normal week, they would put in seventy to eighty hours, pushing against impending deadlines and the boundaries of science. At nights and over the weekends, they answered email, wrote six-page narrative memos, and like their counterparts on the concurrent Alexa and Fire Phone projects, prepared incessantly for regular reviews with Bezos. “We were all living in a cave,” said engineering director Bali Raghavan.

One told me that the original Go store, its adjoining kitchens and data center cost more than $10 million.

In most OP1 sessions, he usually spoke last, not to sway the group with his formidable opinion. But this time, he interjected while Agarwal was still giving his presentation. “You guys are going to fail,” he bluntly told the Indian crew. “I don’t need computer scientists in India. I need cowboys. “Don’t come to me with a plan that assumes I will only make a certain level of investment,” Bezos continued, according to the recollection of two executives who were there. “Tell me how to win. Then tell me how much it costs.”

But who can say what assails the judgment of exorbitantly wealthy and famous men as they enter middle age?

In 2012, AWS introduced Redshift, a so-called data warehouse that allowed companies to analyze the data they stored in the cloud; in 2015, it rolled out Aurora, a relational database. These were typically Amazonian names: geeky, obscure, and endlessly debated inside AWS, since according to an early AWS exec, Bezos had once mused, “You know, the name is about 3 percent of what matters. But sometimes, 3 percent is the difference between winning and losing.”

AWS’s culture was a microcosm of Amazon’s: tough, unrelenting, and focused on meeting impossibly high standards. Jassy and his fellow managers asked searing questions of their underlings and hammered anyone without suitable answers or who didn’t embrace accountability for a problem within their purview. Daily operations were driven by data-filled six-page narratives and the obsessive contemplation of the needs of customers. When employees returned strong results, attention always turned to the ways in which they could have done better. One former executive described the mentality this way: “We were really good at going up to the gold medal podium and complaining that our medals weren’t shiny enough.”

Bezos also conceded that the old process had grown too negative, explaining his sudden appreciation for its flaws to a group of large Amazon investors at a private meeting: “Imagine if you sit down with your wife once a year. You tell her all these great things that you love about her, but then at the end you say, ‘Also, you’re just a little bit fat.’ That’s the only thing that’s going to stick with her from that entire conversation!” As Bezos delivered the punch line, he burst into laughter, according to an investor who was at the meeting: “We want to have a performance review system that doesn’t tell our employees that they’re fat.”

After abandoning a marketing plan to promote the launch by wrapping the entire Empire State Building in gift paper, Amazon introduced Prime Now on December 18, 2014.

They set “S-team goals” such as requiring the team to grow product selection by a certain amount. Around five hundred such goals were established inside Amazon and approved by the leadership committee at the end of every calendar year, establishing the most important metrics for each business unit at the company. Teams that owned those goals were required to supply frequent updates on their progress and explanations if they fell behind schedule. It was a crucial way that the S-team managed a sprawling amalgamation of loosely affiliated business units.

Amazon also sent in the “principal engineers”: an elite squad of about a dozen technical wizards at the company who parachute into troubled projects to diagnose problems. The principal engineers interrogated Treasure Truck employees for two weeks and authored the “correction of error” or COE report, the top-secret document prepared inside Amazon when something goes awry. In the middle of this painful self-examination, disaster struck again: a junior employee accidentally triggered the release of text messages to all the customers who had signed up for Treasure Truck, incorrectly announcing the imminent sale, and the $99 paddleboards.

The Fresh team sent a sample to Bezos’s office, and word came back a few days later that he was satisfied. The project once again represented a different style of innovation within Amazon. Employees didn’t “work backwards” from their idealized customers, who had never asked for such a creation. They worked backwards from Bezos’s intuition and were catering to his sometimes eclectic tastes (literally). Bezos was right a lot, particularly when it came to cutting-edge technology. But in the end, the single cow burger and other culinary innovations introduced within Amazon Fresh generated little buzz or increased business.

Colleagues said that [Dave] Clark had raw management skills and a temper that flared when employees didn’t carefully follow his instructions. He earned a nickname, “the Sniper,” for his proclivity to lurk quietly on the sidelines and identify and fire slacking underlings. He could also be cavalier toward his employees. At all-hands meetings, he would invariably brush aside questions by answering, “I’ll get back to you on that,” but rarely would. Finally, at one meeting, fed-up workers packed the first few rows wearing T-shirts that read, “I’ll get back to you on that.” Clark later insisted that he appreciated the feedback and even kept one of the shirts.

At one memorable meeting to review a team’s proposal to put RFID chips into Amazon pallets and packages, to better track them through the FCs, [Dave] Clark walked in, apparently unimpressed with the plan, and said: “Tell me why I shouldn’t just fire all of you right now.”

Former employees who worked on the app said the development of Rabbit was haphazard and rushed. At first it was missing features that might remind drivers to take breaks or that would choose routes in such a way as to minimize the statistically dangerous act of making a left turn—a key safety feature long known to UPS and FedEx. “Like a lot of things at Amazon, it was ready, fire, aim,” said Trip O’Dell, the former head of the design team that worked on Rabbit. “They just rapidly expanded it and assumed they would fix it later.” ... Issues with the app cascaded, O’Dell said. Routing was poor, and drivers found it difficult to progress from one delivery to another. Fraud was prevalent, with drivers finding loopholes to get paid for work they didn’t do. One problem was that dueling two-pizza teams in Seattle and Austin worked concurrently on versions of Rabbit for iOS and Android, magnifying the confusion among delivery firms. “All of the things were wrong with it all at once,” O’Dell said. “It was not a good app.”

While they would never publicly admit it, Amazon’s senior leaders were happy to operate with more independence, and with fewer of the founder’s impossibly probing questions and demanding ambitions. Meetings with Bezos could still go sideways, resetting projects and depleting employee morale. Even the most inconsequential of utterances from the sagacious chief executive could instigate a flurry of wheel-spinning and white paper–writing inside the company.

In perhaps an indication of Bezos’s initial trepidation about advertising, he had kicked off the effort in the mid-2000s by thinking not about the type of ads the company could accept but the kinds that it shouldn’t. S-team members remember Bezos handing out a list of products that he felt should never be promoted on the site, such as guns, alcohol, online dating sites, dietary supplements, and financial services that pushed people into high-interest loans. The S-team spent hours debating the list and the relative merits of getting into the advertising business.

Though he wasn’t meeting with advertisers, Bezos nevertheless made his presence felt. In the early years of the ad effort, he wanted to review each large campaign, particularly when they ran on the Kindle Fire tablets that debuted in 2011 and featured full-screen color ads. Jeff Blackburn and Paul Kotas, the engineer who managed the technology side of the advertising business at the time, also personally reviewed ad campaigns. Their exacting standards and peculiar aesthetic requirements drove Amazon’s ad execs and their clients crazy. But Blackburn and Kotas had good reason: they didn’t want Amazon to do anything to harm customer trust or interfere with online purchases, at the time the true revenue engine of the company. Often, their reactions consisted simply of a single word: no.

Inside the advertising group, a few of these battles became notorious. Over one holiday, Paul Kotas vetoed the particular hue of blue in ads by the Ford Motor Company, because the display campaign felt “Sunday circular.” Amazon also told the wireless carrier T-Mobile that its trademarked magenta-pink logo was too bright and distracting.

But advertisers came to view the company as arrogant and remote. “We would be met with warm handshakes to start the relationship, and by the end they were tired of us,” said Steve Susi, an Amazon creative director for five years.

Other than a fondness for punk and new wave music, Kotas was, like many of his colleagues, obsessed about metrics, like the length of time it took for ads to load, and fanatical about the leadership principles, such as frugality.

By 2017, revenues from sponsored products had eclipsed those from display advertising like banner ads and would soon after leave them in the dust. That year, sales in the “other” category on Amazon’s income statement (the former home of AWS revenues), where the company parked advertising revenues, hit $4.65 billion—a 58 percent jump from the year before.

In the midst of this realignment, Bezos found another way to reduce fixed costs, flatten his organizational chart, and avoid a specter that he dreaded: that Amazon might become a stodgy “Day 2 company.” He issued a company-wide mandate (“He’s doing that all the time, of course, and people scramble like ants being pounded with a rubber mallet whenever it happens,” wrote former Amazon engineer Steve Yegge in a 2011 blog post). From henceforth, all Amazon managers—whose direct reports consisted primarily of other managers—would have to have a minimum of six employees reporting to them. Though it sounded innocuous, the directive, dubbed “span of control,” set off the equivalent of a neutron bomb inside the company. Senior managers with only three, four, or five direct reports had to reach into their organizations and appropriate employees from a subordinate to get to six direct reports, leaving the underling without the necessary number.

Bezos was aware of these [local charity] initiatives, said Amazon employees who worked on them. He supported them, several felt, because they boosted Amazon’s image and required a relatively minor investment of dollars and his own time. He was characteristically focused on the business and largely transactional when discussing community involvement. Internal documents at the company advocated that Amazon do enough to maintain its “social license to operate”—the business concept that refers to the public’s acceptance of a company, its employees, and business practices.

By 2018, thousands of new hires were joining the company each month. Employees working in office buildings like Doppler and Day 1 recalled that the offices were so cramped, coworkers were sometimes given desks in hallways. Company events were packed, and that year was the first that Amazon wasn’t able to hold the annual summer picnic at CenturyLink Field. Now there were simply too many employees. Recruiting was also becoming more difficult. The company was exhausting the number of people—engineers, but also lawyers, economists, and human resource executives—it could convince to move to the escarpment of fog and rain clouds that hung over the Pacific Northwest.

Philadelphia was not a hotbed of engineering talent, and in the meeting, AWS chief Andy Jassy, according to one person’s recollection, opined that he disliked the city, which was the bitter rival of his favorite football team, the New York Giants, and suggested that he and his employees would never want to live there. Jassy was apparently joking, but some members of the HQ2 team, coming off months of detailed, quantitative work, later expressed exasperation that the process was now exposed to the arbitrary personal preferences of senior executives.

After the head-tax proposal, Bezos contacted John Schoettler and ordered the real estate division to stop construction on “Block 18,” a seventeen-story tower near Day 1, and to sublease most of the 800,000-square-foot building that Amazon had completed at nearby Rainier Square rather than occupy it. The real estate team predicted that the move would cost the company more than $100 million, according to a person familiar with its calculations (although this person said the company later broke even on the transaction). But Bezos said he didn’t care: Amazon wasn’t going to grow in a city that didn’t want it. At the same time, Bezos instituted another internal edict: he capped Amazon’s Seattle headcount at around fifty thousand employees. Amazon, which already occupied more than 19 percent of prime office space in the city, was due to hit that number within twelve months. After that point, managers inside the company would have to funnel their headcount growth to Amazon offices in other cities. Schoettler and the real estate team scrambled to accommodate the new demand. Since Amazon employed around seven hundred people only fifteen minutes away, across Lake Washington in Bellevue—an affluent Seattle bedroom suburb that at the time was opportunistically running promotional campaigns targeting local corporations—Amazon execs decided that the Seattle overflow could go there, and established a target of moving twenty thousand employees. That fall, Amazon would sign a lease in Bellevue for the twenty-story former headquarters of online travel company Expedia.

“There is no company in the world that is more complex and difficult for outsiders to understand than Amazon,” said Kurt Zumwalt, Amazon’s treasurer for fifteen years, before he left in 2019. “This is not a typical corporate conglomerate like Berkshire Hathaway or General Electric. Almost every aspect of Amazon is built around subtly increasing its connection to customers. The power of the business model is the combination of the sum of its self-reinforcing businesses and services, enabled by world-class technology, operational excellence, and a rigorous review and measurement process.”