Bezos’s 2016 Shareholder Letter: Four Essentials for Day 1

Robin: “If you only have time for a few of Bezos’s shareholder letters, do not miss 2016. It discusses Day 1 and Day 2, ideas now passed from person to person throughout business: remember your original purpose, keep innovating, and move forward.” #BezosShareholderLetter2016 It’s All About The Long Term. To our shareholders: “Jeff, what does Day 2 look like?” Someone asked me that at a recent meeting. For decades I have reminded people that it is Day 1. I work in an Amazon building called Day 1, and when I move buildings, I take the name with me. I have spent time thinking about this question. “Day 2 is stasis, then irrelevance, then suffering, then death. That is why we remain in Day 1.” Of course, the change happens slowly. A company may spend a very long time in Day 2, but its ultimate fate will always arrive. What interests me is how to refuse Day 2. What techniques and methods help preserve Day 1’s vitality, even within a very large organization? There is no simple answer. There are many elements, paths, and traps. I do not know all the answers, but I know some. This is the starter kit for defending Day 1: customer obsession, skepticism toward proxies, eager adoption of external trends, and high-velocity decision-making. /Customer obsession/ A business can center on many things: competitors, products, technology, business models, or something else. But to my mind, customer obsession is the most effective way to preserve Day 1’s vitality. Why? Customer focus has many benefits, but the biggest is that customers are never satisfied—“customers are always beautifully, wonderfully dissatisfied”—even when they say they are happy and you are doing wonderfully. Even without realizing it, they always want better service. Your commitment to delighting them leads you to innovate on their behalf. No customer told Amazon they needed Prime, but once it existed, they discovered it was exactly what they wanted. I could give countless examples. Remaining in Day 1 requires patient experimentation, accepting failure, planting seeds, protecting seedlings, and doubling down once customers are delighted. A customer-obsessed culture provides the best soil for all this to happen. /Resist proxies/ As companies grow larger and more complex, they must address proxies. These appear in all shapes and sizes and are dangerous, subtle, and very Day 2. A common example is process becoming a proxy. Good processes serve you and your customers. But if you are not careful, the process itself becomes the problem. This happens easily in large organizations: process becomes a proxy for the result you seek. You stop examining actual outcomes and merely focus on compliance with the process. You often hear a junior leader defend a poor outcome: “Well, I followed the process.” A more experienced leader investigates and improves it. Process is not the real issue. We must keep asking whether we control the process or it controls us. In a Day 2 company, you may find the latter. Another example: marketing and customer research can become proxies for actual contact with customers, which is especially dangerous when innovating and designing products. “55% of beta testers are satisfied, up from 47% in the first test.” Such results are difficult to interpret and may unintentionally mislead. Good innovators and designers understand customers deeply. They invest tremendous energy in developing intuition. They study and understand individual stories rather than staring at a few survey numbers. They live with their designs. I am not against beta tests or research. But you, the product or service owner, must understand customers, have a vision, and love what you offer. Only then can testing and research uncover blind spots. A remarkable customer experience starts with heart, intuition, curiosity, playfulness, courage, and taste. None of that appears in a survey. /Embrace external trends/ If you cannot adopt emerging trends, the outside world will push you into Day 2. Fighting them means fighting the future. Embrace them and gain a tailwind. These big trends are not particularly difficult to spot—people discuss them everywhere. Strangely, however, large companies find them hard to embrace. We are in the midst of an obvious one: machine learning and artificial intelligence. For decades, computers have automated tasks programmers could express through clear logic and algorithms. Current machine-learning techniques let us automate tasks whose rules are much harder to describe. At Amazon, we have researched machine-learning applications for years. Some are obvious: Prime delivery drones, Amazon Go stores without checkout lines, and Alexa, a cloud-based AI assistant. More applications lie beneath the surface. We use machine-learning algorithms to forecast order volumes, rank product searches, recommend products, place merchandise, detect fraud, translate, and more. Less visible though they are, most machine-learning effects take this form—quietly changing our core operations. Within AWS, machine learning and AI have sharply reduced costs, allowing companies of all sizes to benefit from our services. Using our prepackaged deep-learning frameworks on P2 compute, customers can build powerful applications ranging from disease detection to improving corn yields. We have also made higher-level Amazon services easier to use. Amazon Lex, at Alexa’s core, Amazon Polly, and Amazon Rekognition remove the heavy lifting in natural-language processing, speech generation, and image analysis. Simple API calls make them accessible without machine-learning expertise. Watch this area; much more is coming. /High-velocity decision-making/ Day 2 companies make high-quality decisions, but very slowly. Preserving Day 1’s energy and vitality requires decisions that are both high quality and fast. That is easy for startups and difficult for large organizations. Amazon’s senior managers are trying to keep decisions fast. Speed matters in business, and a fast-moving decision environment is also more enjoyable. We do not have every answer, but here are some ideas. First, never impose a one-size-fits-all decision process. Many decisions are reversible and can use a lighter process. For anyone asking, “What if we are wrong?”, see Bezos’s 2015 shareholder letter. Second, most decisions should be made once you have about 70% of the information. Waiting for 90% is too slow. Regardless, you must recognize and correct errors quickly. If you are good at adjusting course, being wrong costs less than you imagine, while being slow is certainly expensive. Third, use “disagree and commit” more often. It saves considerable time. If you believe in a direction but the team cannot agree, it helps to say: “Look, I know we disagree. Will you take a chance with me? Disagree and commit?” No one can know the exact answer then, and you will probably get a quick yes. This is not one-way. If you are the boss, you must do it too. I disagree and commit all the time. We recently started filming an Amazon Studios original series. I told the team my concerns: whether it was interesting enough, whether production would be too difficult, the unfavorable commercial outlook, and our other opportunities. They saw it differently and wanted to proceed. I immediately replied: “I retain my reservations but support you proceeding, and hope it becomes our most popular series.” Imagine how much slower we would have been if they had needed to persuade me rather than receive my commitment. Notice what this example does not mean: I was not thinking, “These people have no idea what they are doing, but I have no time to chase them.” It was a genuine disagreement, an honest expression of my view, a chance for the team to consider it, and a quick, sincere commitment to their decision. Given that they have already brought home 11 Emmys,6 Golden Globes, and 3 Oscars, I am happy they let me into the room at all! Fourth, identify misalignment quickly and escalate it. Different teams sometimes have different objectives and perspectives. They are misaligned, and no amount of discussion or meetings can resolve it. Without escalation, the default conflict-resolution mechanism becomes exhaustion. Whoever outlasts the other wins. I have seen many sincere disagreements over the years. When we decided to introduce third-party sellers, many well-intentioned employees disagreed with management. That major decision contained hundreds of smaller decisions, many requiring coordination by senior management. “I cannot outlast you” is a terrible decision process: lengthy and exhausting. Escalating problems quickly is better. Are you concerned only with decision quality, or also decision speed? Are global trends giving you a tailwind? Have you replaced important things with proxies? Most importantly, are you delighting customers? We can combine a large company’s capabilities with a small company’s spirit, but we must actively choose to do so. As always, I have attached the 1997 letter. Still Day 1! Jeffrey P. Bezos (Translated into Chinese by Vinchent)