Bezos’s 2018 Shareholder Letter: Intuition, Curiosity, and the Power of Wandering

#BezosShareholderLetter2018 How do you put customers at the center? With heartfelt intuition, imagination, and innovation. As your company grows, you also have to increase the scale of the failures you are willing to risk through experimentation. To our shareholders: Something worth noticing has happened over the past 20 years. Look at these numbers: 1999 3% 2000 3% 2001 6% 2002 17% … (omitted) 2015 51% 2016 54% 2017 56% 2018 58% These percentages show the proportion of sales made by third-party sellers, most of them small and medium-sized businesses. Over 20 years, that share rose from 3% to 58%. Put simply: Third-party sellers keep giving Amazon’s own retail business a resounding slap in the face. That is more formidable than you might imagine, because Amazon’s first-party business has also grown enormously over those years, from US$1.6 billion to US$117 billion, a compound annual growth rate of 25%. Third-party sellers, meanwhile, grew from US$100 million to US$160 billion, an annual growth rate of 52%. For an external comparison, eBay grew at 20% a year during that period, from US$2.8 billion to US$95 billion. Why did independent merchants sell better on Amazon than on eBay? And why could they grow faster even against Amazon’s powerful first-party business? There is no single answer, but we know one important part: We helped third-party sellers beat our own retail business by building the best selling tools we could imagine. There are many: tools for inventory management, payment processing, shipment tracking, reporting, and cross-border sales. We create more every year. But the most important are FBA and the Prime membership program. Together, these two initiatives substantially improved the experience of buying from third-party sellers. Both now look successful, but people did not see them that way when we launched. We kept investing despite significant financial risk and internal objections. Over the years, we have also continued investing heavily in other ideas. We could not predict these programs’ eventual form, let alone know whether they would succeed. But we pushed them forward with intuition and heart, and continued to believe in them. /Intuition, Curiosity, and the Power of Wandering/ Early in Amazon’s life, we knew we wanted a culture of builders: curious people who love exploring. They should enjoy innovation and retain a beginner’s curiosity even after becoming experts. This builder’s mindset gives us a humble conviction when facing big, difficult opportunities: create, try, recreate, try again, and keep repeating. Eventually, we can make it work. Sometimes you know what you are doing and can be highly efficient: set a goal and execute. Wandering, by contrast, is not especially efficient. But it is not random wandering either. It is guided by intuition, curiosity, and a customer-centered spirit. Even if the road is long, the journey feels worthwhile. Wandering is an important counterbalance to efficiency. You need both. Disruptive, “nonlinear” innovations come from wandering. AWS’s millions of customers range from startups and large enterprises to governments and nonprofits. Each is building better solutions for its own customers. We spend a great deal of time considering what these organizations and the people in them want: developers, development managers, operations managers, CIOs, chief digital officers, chief information security officers, and so on. Much of our work at AWS is based on listening to customers. It is essential to ask customers what they want, listen carefully, and devise a plan to deliver it well and quickly. No business can keep going without this kind of customer focus. But it is not enough. The greatest opportunities concern needs customers do not yet recognize. We must put ourselves in their position and invent for those needs. That requires imagination. AWS itself is an example. Nobody was asking for AWS. Yet the world did need something like it; people simply did not know it. We had an intuition, followed our curiosity, took the necessary financial risks, and began building—reworking, experimenting, and iterating countless times. The same pattern has appeared many times within AWS. We built DynamoDB, for example, a highly scalable, low-latency key-value database now serving thousands of customers. On the listening side, we heard customers complain that commercial databases were expensive and difficult to use, so we built Amazon Aurora. We are also confident about specialized databases for specialized domains. Over the past 20 to 30 years, companies have generally used relational databases to run their services. Developers’ familiarity with these databases made them the default choice. They were not ideal, but data volumes were relatively small and latency requirements relatively loose, so they were workable. Today, however, many applications require real-time access to huge volumes of data, from terabytes to petabytes, with low latency and millions of requests per second. The options now include not just DynamoDB, but in-memory databases such as Amazon ElastiCache, time-series databases such as Amazon Timestream, and ledger solutions such as Amazon Quantum Ledger Database. Using the right tool for the job saves money and improves efficiency. We are also working on helping companies use machine learning more effectively. We have been doing this for a long time. As with other innovations, our first attempts to expose machine-learning tools through usable interfaces failed. After years of wandering—experimentation, iteration, refinement, and listening to customers—we finally developed SageMaker. Launched 18 months ago, SageMaker removes complex concepts and details from each step of machine learning, making AI accessible to everyone. Thousands of customers now build machine-learning applications with AWS SageMaker. We will keep upgrading the service, including new reinforcement-learning capabilities. Reinforcement learning has a steep learning curve and many moving parts; its difficulty means only a few large companies can fully understand and apply it. These applications would not exist without the spirit of inventing for customers, curiosity, and perseverance. Customers are responding: AWS now brings in 3 million in annual revenue and is growing rapidly. /Imagining the Impossible/ Amazon still holds only a small share of retail today—a single-digit percentage. In every country where we operate, there are retailers bigger than us. That is largely because 90% of retail remains offline, in cold brick-and-mortar buildings. We spent years thinking about how to serve physical retailers, but felt we first needed to invent something that would delight customers in a physical setting. That led to the cashierless Amazon Go store and a clear goal: eliminate queues. Nobody likes standing in line. You should be able to enter, take what you want, and leave. Getting there was extremely difficult, particularly technically. It required hundreds of smart, skilled computer scientists and engineers around the world. We had to build proprietary cameras and shelves and develop new computer-vision algorithms, including ways to combine images from hundreds of cameras. The technology also had to be invisible, without disrupting shopping. Customers love the result and describe shopping at Amazon Go as “magical.” We now have 10 stores in Chicago, San Francisco, and Seattle, with more to come. /Failure Must Scale Too/ As a company grows, everything must scale, including the scale of failure. If your failures are not getting bigger, you cannot innovate meaningfully at your present size. Amazon will experiment in ways appropriate to our size. If an experiment could produce a multibillion-dollar failure, we certainly will not enter it recklessly. We will work hard to make good bets, but not every bet we make will succeed. The ability to take these high risks is a benefit that large companies like ours can provide to customers and society. The good news for shareholders is that one success can pay for many past failures. Development of the Fire phone and Echo began at roughly the same time. When the Fire phone failed, we quickly transferred what we had learned—and the developers—to Echo and Alexa, accelerating development. The original vision for Echo and Alexa came from the computer in Star Wars, and from two fields in which we had spent years wandering: machine learning and cloud computing. Early at Amazon, AI algorithms were mainly used for product recommendations, while AWS gave us a strong cloud foundation. After years of development, Echo launched in 2014, supported by Alexa running on AWS. No customer was clamoring for Echo. It was unquestionably a product of our wandering. Market research would not have helped. Imagine asking people in 2013: “Would you like a cylindrical black box about the size of a small Pringles can, sitting in your kitchen, talking to you and playing music?” I guarantee they would look at you as if you were an idiot and say, “No thanks, I’ve got something to do. I’m leaving…” Since the first-generation Echo, customers have bought over 10 million Alexa devices. Last year, we improved Alexa’s ability to understand requests and answer questions by 20%. We also added hundreds of thousands of facts to make it more knowledgeable. Developers doubled Alexa’s skills to more than 80,000. Customers interacted with Alexa over 1 million more times in 2018 than in 2017. More than 150 products now have Alexa built in, from headphones and PCs to cars and smart-home devices. More devices are on the way! One final point before closing. As I said in our first shareholder letter more than 20 years ago, we are committed to recruiting and retaining talented people who think like owners. That requires substantial investment in employees. As with Amazon’s other businesses, we rely not only on data analysis but also on intuition and heart to find our direction. Last year, we raised our minimum wage to US$15 per hour for every full-time, part-time, temporary, and seasonal employee across the United States. The increase benefited more than 250,000 Amazon employees and 100,000 seasonal workers employed across the country during the previous Christmas holiday. We firmly believe investing in employees benefits our business. But that was not the main reason for the decision. Our wages were already competitive. This time, we wanted to lead, offering pay well above merely “competitive.” We did it because we believed it was right. Today, I challenge our biggest retail competitors—yes, you know who you are! Raise your employee benefits to our level and your minimum wage to US$15, or even US$16, and put the ball back in our court. This is competition that benefits everyone. Our other employee programs come from the heart too. I have previously mentioned Career Choice, which reimburses 95% of the cost of whatever employees want to study. More than 16,000 employees have taken advantage of it, and that number keeps growing. Similarly, Career Skills helps hourly workers learn essential workplace skills such as writing a resume, communicating, and using basic computer functions. Last October, extending that commitment, we signed the President’s Pledge to America’s Workers and announced that at least 50,000 U.S. workers would improve their skills through our innovative training programs. Our investments are not limited to current employees. To train the future workforce, we pledged 50 million to our recently announced “Amazon Future Engineer” program, which aims to improve STEM and SC education for elementary, secondary, and university students across the United States, particularly by attracting girls and “minorities” to these fields. Our veteran-recruitment program continues too, and we are close to our goal of hiring 25,000 veterans and family members by 2021. Through Amazon Technical Veterans Apprenticeship, we provide cloud-computing training so veterans can adapt to Amazon and move quickly into their roles. A huge thank-you to our customers for holding us to higher standards as we serve you; to shareholders for their continuing support; and to employees for their hard work and determination. All Amazonians are listening to customers and continually inventing from their perspective. As usual, the letter is attached. Day 1. Jeffrey P. Bezos (Chinese translation by Vinchent) ——Verification note—— Source: Amazon’s 2018 shareholder letter, an official public document. The Chinese translation was reposted from Dedao App user Vinchent, as credited in the original post. Verification: Compared with the official original in Amazon’s official shareholder-letter collection, cross-checked against SEC documents/official PDFs. Selected major figures and facts were consistent; the entire text was not compared word for word. Corrections (four differences between the Chinese translation and the original): • “3 million in annual revenue” should be a US$30 billion annual revenue run rate. • “Over 10 million Alexa devices” should be over 100 million. • “Hundreds of thousands of facts added” should be billions of facts. • “Over 1 million additional interactions” should be tens of billions more interactions.