Author | Amazon CEO Andy Jesse
Translate | Lu Anqi
How amazon after the epidemic achieved the equivalent of three years of growth in 15 months.
Recently, Amazon (AMZN.US) CEO Andy Jassy published his first annual shareholder letter since he took over the seal from founder Jeff Bezos last year.
Dear Shareholders,
Over the past 25 years at Amazon, I've had the opportunity to write many emails, letters, and keynotes for employees, customers, and partners.
However, this is the first time I've had the privilege of writing an annual shareholder letter as Ceo of Amazon. Jeff set a very high standard for these letters, and I will try to make them worth reading.
When the epidemic first began to spread in early 2020, few thought it would develop into its current size and continue for a long time. Until then, no matter what role Amazon has played in the world, this image will be further amplified.
Because most premises will be closed for a long time, people will have to work from home. It also means that hundreds of millions of people rely on Amazon online to buy fixtures, food, clothes and a variety of things to help them through this unprecedented predicament.
Businesses and governments have had to shift from working remotely, almost overnight, working with colleagues, using in-house technology, and so on. Amazon Web Services (AWS) plays an important role in enabling business continuity.
Whether companies are seeing an extraordinary surge in demand or demand shrinking rapidly as external consumption decreases, the elasticity of this cloud service to rapidly scale capacity up and down, and the extraordinaryly wide range of capabilities of the Amazon network, have helped millions of companies adapt to these difficult environments.
The pandemic presents both challenges and opportunities
During the pandemic, the demand trajectories of our Amazon web services and consumer businesses have been different. In the first year of the pandemic, Amazon Web Services revenue continued to grow rapidly — 30% year-on-year in 2020 ($35 billion in 2019), but at a slower pace than 37% in 2019.
In part, this is because many businesses are experiencing uncertainty and slowing demand, but also because we are helping enterprises optimize their AWS layout to save money.
At the same time, companies are taking a step back and deciding what to change about the pandemic.
Many conclude that they don't want to continue managing the technology infrastructure themselves and decide to accelerate their migration to the cloud. With the economic recovery, the joint efforts of many companies have allowed AWS's revenue growth rate to return to 37% year-on-year in 2021.
Instead, our consumer revenues grew significantly in 2020. In 2020, Amazon's North American and international consumer revenue grew 39% year-over-year (on a revenue base of $245 billion in 2019), and this significant increase continued into 2021.
In the first quarter of 2021, revenue increased by 43% year-over-year. This volume is staggering, as we achieved the equivalent of the 3-year growth we had originally expected in about 15 months.
The world reopened from the end of the second quarter of 2021, with more and more people venturing out to eat, shop and travel, and consumer spending redistributing across more entities.
We're not sure what's going to happen in 2022, but the fact that we continue to grow at double-digit rates (29% consumer CAGR over two years) is encouraging as customers appreciate the role Amazon has played for them during the pandemic and are starting to use Amazon for more home purchases.
This growth also comes with short-term logistical and cost challenges. We spent the first 25 years of Amazon building a very large delivery network and then had to triple it in the last 24 months to meet new customer demands.
When we brought this new capacity online, the labor market was quite tight, which made it extremely difficult for suppliers and sellers to send all their inventory and place it closest to their customers.
This, coupled with the fact that sea, air and trucking capacity is becoming scarce and expensive, creates additional transportation and production costs.
Supply chains have been disrupted like never before. We had hoped that as 2021 drew to a close, the main impact of COVID-19 would fade. But then the Omiljung strain began to emerge in December, which also had a global impact, hitting people's ability to work hard.
In late February, fuel costs and inflation became bigger issues to deal with as the Russian-Ukrainian conflict followed.
As a result, 2021 is a crazy and unpredictable year, continuing the trend since 2020. But I remain proud of the incredible commitment and effort of our employees around the world.
Without the dedication and extraordinary effort of the team, I'm not sure any of us would have weathered the pandemic in the same way. I will always be grateful.
For companies of any size, it is unusual to be able to respond to events as discontinuous and unpredictable as an outbreak.
What makes it possible for Amazon to do this? Because we didn't start from a starting point. For nearly two decades, we have been iterating and transforming our fulfillment capabilities. In every business, we are constantly experimenting and inventing.
We are never complacent about customer experiences, we believe that these customer experiences are always better, and we strive every day to make our customers' lives better and easier.
The beauty of this task is that you never run off the runway; clients always want something better, and our job is to listen to their feedback, imagine what else is possible, and invent on their behalf.
It's often assumed that game-changing inventions just bubble up from someone's head — a light bulb goes out, a team executes the idea, and then you have a new invention, which is a long-term breakthrough success — but that rarely happens.
A little-known fact about innovative companies like Amazon is that we relentlessly debate, redefine, tinker, iterate, and experiment to get the seeds of a great idea and turn it into something that resonates with our customers and, over a long period of time, meaningfully improve the customer experience.
Next, let me give you some examples of Amazon.
Today's Amazon: A list of core products
Our fulfillment network: Back in the early days of the pandemic, it was almost impossible for us to start working in March 2020 and meet anything close to the needs of our customers. For 20 years, we've been innovating and iterating fulfillment networks, trying to reduce the time it takes to get goods into the hands of our customers.
In the early 2000s, it took an average of 18 hours for a commodity to pass through the fulfillment center and be delivered to a suitable truck. Now, it only takes 2 hours. The network is reliable and cost-effective, and offers additional services for Amazon Elite members who ship within a few days.
It took us several years to build a huge fulfillment center, strong logistics and transportation capabilities, and realign almost all processes in the facility.
In the long run, in 2004 Amazon had seven fulfillment centers in the United States and four in the rest of the world. There are currently no additional distribution stations that connect our fulfillment and sorting centers to "last mile" delivery trucks.
By the end of 2021, we have 253 fulfillment centers, 110 sorting centers and 467 delivery stations in North America, and 157 fulfillment centers, 58 sorting centers and 588 delivery stations worldwide.
The delivery network has more than 260,000 drivers worldwide and Amazon Airlines cargo fleet of more than 100 aircraft. Over the past fifteen years, this represents more than $100 billion in capital investment and more than 1 million Amazonians, through countless iterations and small-scale process improvements.
Ironically, before the COVID-19 pandemic began, we decided to invest billions of dollars in a few years to deliver more and more major goods in one day. This plan has been slow to progress due to the challenges of the pandemic, but we have resumed our focus on these.
Delivering a large number of goods in one day is difficult (especially millions of items) and initially expensive because of the infrastructure to build to scale this effectively. But we believe that more than 200 million preferred customers will tell you very clearly: the sooner the better. They'll love that.
So, the ability to ship millions of items in a matter of days (and more often in a day) doesn't come from a moment, nor does it develop in a year or two.
This didn't come easily because we put ourselves in the shoes of our customers, knew what they wanted, organized Amazonians to work together to invent better solutions, and invested a lot of money and human resources in 20 years (often much earlier than the time to payback).
This type of iterative innovation never ends, but leads to a better long-term customer experience, customer loyalty, and shareholder returns.
Amazon Web Service: When we define Amazon Web Services and work backwards on the services, we constantly raise the boundaries of one of the biggest core points in product development – where the functional boundaries of V1 are.
In particular, an early meeting of the Core Compute Services Elastic Compute Cloud ("EC2") was scheduled for one hour and took three hours as we enthusiastically debated whether computing services could be launched without a block storage companion (a form of network-attached storage).
Everyone agrees that this is important for a complete computing service; however, it takes a year to get one ready. The question becomes: Can they provide useful services to customers and meaningful value before they have all the features they want?
We believe that if we listen to our customers and iterate quickly, the initial launch of EC2 may lack features. This approach works if customers do iterate quickly; but if not, it would be disastrous.
We launched EC2 in 2006, a data center in a certain part of the world with only one instance size, using only the Linux operating system (no Windows), no monitoring, load balancing, automatic scaling, or, yes, no persistent storage.
EC2 was initially successful, but it was far from becoming a multibillion-dollar service until we added the missing features listed above.
In the early days of Amazon Web Services, people sometimes asked us why computing wasn't just an undifferentiated commodity.
However, there is a lot more to compute than servers. Customers require various types of computing (e.g., server configurations optimized for storage, memory, high-performance computing, graphics rendering, machine learning), multiple form factors (e.g., fixed-size, portable containers, serverless functionality), persistent storage of various sizes and optimizations, and a range of networking capabilities. Then, install a CPU chip in your computer.
The industry has been using Intel or AMDx86 processors for years. We have important partnerships with these companies, but we realize that if customers want to further increase price and performance, they must develop their own chips.
Our first general-purpose chip was the Graviton announced in 2018. This helps a subset of workloads run more cost-effectively.
However, it wasn't until 2020, after learning from Graviton's experience and innovating with the new chip, that the Graviton2 chip delivered an extraordinary performance, which was 40% better than the latest generation of x86 processors. Think about how much of an impact a 40% increase in computing performance will have.
Computers are used in every technology. This is a big deal for the client. In addition, while Graviton2 has been a huge success to date (48 of Amazon Web Services EC2's top 50 customers have adopted it), the chip team announced Graviton3 last December (offering a 25% improvement on the basis of Graviton2's relative earnings).
At EC2 (Amazon Web Services), the list of products invented and delivered for customers is incredible. Iterative innovation not only provides customers with more Amazon Web Services features than anywhere else (an important distinction), but also allows us to deliver more game-changing products.
Device-related: Amazon first ventured into devices when it released the Kindle in 2007. It's not our most complex industrial design (it's milky white, some people hold it uncomfortable), but it's revolutionary.
Because at the time, customers could download any of the more than 90,000 books (now millions) in 60 seconds — we got better and faster at building attractive designs.
Shortly thereafter, Amazon launched tablets, then phones (with front-facing cameras and gyroscopes that offer dynamic viewing angles and various 3D experiences).
Although the latter two projects may have been too late to succeed and shifted those resources elsewhere, we hired some great long-term builders and learned valuable lessons from this failure — lessons that have been well implemented in devices like the Echo and FireTV.
When I think about the first Echo device — and what Alexa could do for customers at the time — it's noteworthy, but it's far less than what teams might be able to do today.
Today, there are hundreds of millions of Alexa-enabled devices around the world (at home, in the office, in the car, in hotel rooms, Amazon Echo devices, and third-party manufacturer devices); listening to music or watching videos; controlling lights, automating the home; and creating schedules like "Start My Day."
Alexa can tell you the weather, predict commute times based on current traffic conditions, and then broadcast the news; it can also easily order retail items on Amazon; get general or customized news, updates on sporting events and related statistics – we knew early on what Alexa and Alexa-related devices can do for customers, and our goal is to make Alexa the most helpful and resourceful personal assistant in the world, making people's lives easier and better.
We have a lot more invention and iteration to do, and there are several other devices at different stages of development (for example, Ring and Blink offer leading digital home security solutions, and Astro is a brand new home robot to be launched at the end of 2021).
But it's safe to say that every device, whether it's a Kindle, FireTV, Alexa/Echo, Ring, Blink, or Astro, is an invention in progress, and there will be more in the future that will continue to improve the lives of our customers.
Preferred Video: We've been launching a product called Amazon Unbox since 2006, and customers can download about a thousand movies from Head Movies. This makes sense because the bandwidth was slower at the time (it takes an hour to download a video).
However, as homes and mobile devices gain faster bandwidth, and the advent of connected TVs, streaming will become a better customer solution, and we're focusing our efforts on streaming.
In 2011, as part of Amazon's Preferred Subscription, we began offering more than 5,000 streaming movies and shows. Initially, all the content was produced by other studios and entertainment companies. These transactions are expensive, vary from country to country and are only available for a limited time.
So, in order to expand the selection, we started creating our own original shows. Our early shows included short live shows like Alpha House and Betas. Before we announced our first winning series, we also contracted franchises for years of hit Series such as The Boys, Bosch, Jack Ryan, and more.
Along the way, how to create engaging entertainment with memorable moments, how to use machine learning, and other creative techniques to deliver high-quality streaming experiences (with just a click of a mouse, you can get data about actors, TV shows, movies, music, or sports statistics).
You've probably seen this in the recent hit new series Reacher, and hopefully you'll see it in the upcoming Lord of the Rings series, which is expected to be released on Labor Day in 2022.
We also hope that when Thursday Night Football is launched in September 2022, people will see this iterative innovation. It was the first weekly prime-time pure streaming broadcast by the National Football League (NFL) and was exclusively broadcast by Amazon Preferred Video.
Our agreement with the NFL is 11 years. Over the next few years, we will work tirelessly to reinvent the NFL viewing experience for football fans.
Such a frequent record of innovation is not only the reason why more and more sports organizations choose to cooperate with Preferred Video, but also why so many large entertainment companies have become Preferred Channel partners.
Channels enable entertainment companies to leverage the unique technology and viewing experience of preferred videos, as well as a membership base, to offer monthly subscriptions.
Explore companies like Warner Bros., Paramount, Starz, Corus Entertainment, and Globo that they're driving significant membership growth through channels, as well as a better customer experience.
While the Preferred Video Series has made tremendous strides from its inception to the present, over the next 15 years, inventions will surpass the sum of the past. And our team will be passionately committed to providing our clients with the broadest and most compelling collection of content around the world.
Amazon's business philosophy
This same type of iterative invention can be applied to support the efforts of people and communities. Last summer, we added two new leadership principles: striving to be the best employer on the planet, and the greater responsibility that comes with success and scale.
These concepts have always been "invisible" at Amazon, but clear leadership principles help us ask ourselves — and ask more Amazon employees at all levels — whether we're following them.
For example, more than 1 million Amazonians work in our fulfillment network. In 2018, we offered a minimum wage of $15 an hour on average (more than double the federal minimum wage), but it didn't stop there. After continuing to increase salaries, the current hourly wage is more than $18.
In addition to this subsidy, we offer very robust benefits, including comprehensive health insurance, a 401K plan, up to 20 weeks of parental leave, and full tuition insurance for employees who wish to get a college education (whether they remain with our company or not).
Improving the lives of employees is far from complete. We've researched and created a list of the top 100 employee experience pain points and are systematically addressing them.
We are also passionate about further improving the security of our fulfillment network, with a focus on reducing strain, sprains, falls and repetitive stress injuries.
Injury rate is a number that can sometimes be misunderstood. We have operations that are suitable for both "warehousing" and "express and distribution" categories.
In recent U.S. public data, Amazon's recordable incident rate is slightly higher than the average for its warehousing peers (6.4 vs. 5.5) and slightly lower than the average for its express and delivery peers (7.6 vs. 9.1). Although the year-on-year is on average, "the best in the class" is what we pursue.
When I first started my new role, I spent a lot of time on the fulfillment center and the security team, hoping that there would be a winning "silver bullet" that would quickly improve the numbers, but I didn't find it.
At its current scale (Amazon will hire more than 300,000 people in 2021 alone, many of them new and in need of training), rigorous analysis, thoughtful problem solving, and a willingness to innovate are needed to get to where you want to be.
We're dissecting each process path to see how we can improve it further. There are various programs in progress (e.g., rotation programs that help employees avoid spending too much time doing repetitive actions; wearable devices that alert employees when they act in dangerous ways, improved shoes that better protect their toes, training programs on body mechanics, health, and safety practices, etc.).
However, we still have some way to go. Treat it like any other customer experience – keep learning, inventing, and iterating until there are more transformational outcomes. We won't be satisfied until the day we do it.
Future outlook
Similarly, at our current scale, we have left a significant carbon footprint. This is also a big reason for the establishment of the Climate Pledge a few years ago ( a decade ahead of the Paris Agreement , which pledged to achieve carbon-zero emissions by 2040 ).
Significant progress is being made in this regard (committed to providing 100% renewable energy to operations by 2025 – five years ahead of the original 2030 target). We have ordered more than 100,000 electric vehicles to deliver packages, and more than 300 companies have joined our climate commitments).
Given the diversity and density of Amazon's business, which includes shipping billions of packages each year, the challenges we face are different from most companies. We are committed to meeting the challenges, but this requires relentless innovation.
We strive to increase the number of affordable homes in communities where a lot of transactions are fulfilled. The Housing Equity Fund, established a year ago, has exceeded $2 billion.
$1.2 billion has been allocated for affordable housing projects in the Puget Sound area of Washington, Arlington, Virginia, and Nashville, Tennessee.
The last simple example is Kuiper. This is our low-orbit satellite network, and we're going to spend more than $10 billion to build it over the next few years.
Kuiper will provide customers with the least used, or even no fixed broadband connection, changing the way many communities access information and resources (analysts estimate that about 300-400 million customers worldwide fall into this category).
We are optimistic that we also have a really good business model, but we will see that as its features continue to evolve, it will be a real game changer for many years for families and businesses with service needs.
This type of iterative innovation pervades every team at Amazon. I can give similar examples of advertising, groceries, games, Amazon music, telemedicine, or pharmacies.
These stories continue to be written as we experiment quickly, learn, and continue to work to make the customer experience better every day.
If this approach sounds appealing, then a natural question is: What is it takes to be proficient in it? Easier said than done, here are some helpful parts:
1. Hire the right builders: We don't pay attention to proportions when it comes to hiring builders.
We think builders are people who love to invent, who focus on the customer experience, dissect where it doesn't fit, and seek to reinvent it. We need to constantly ask people who "why can't we do it", need people who like to experiment and tinker, and the launch of a product is the starting line rather than the finish line.
2. Plan builders as separable and autonomous teams: It's hard for teams to get deep into what customers care about in multiple areas.
When competing with more mature businesses for resources, it's also difficult to spend enough time on new initiatives; winning bets usually win. Single-threaded teams will better understand customer needs, spend all their time developing, and adapt to the right environment and speed to keep iterating fast.
3. Give the team the right tools and permission to act quickly: speed is not predetermined. It was the captain's choice. You can't wake up one day and start moving fast.
It needs to have the right tools to experiment and build quickly (which is the main reason we created Amazon Web Services), allow teams to make two-way decisions on their own, and set speed expectations.
It's true that speed is extremely important for every stage of every business. Those who are slower than their competitors will fall behind over time.
4. Blind faith, but no false hope: This is a lyric from one of my favorite Spitfire song, "Congregation."
When invented, you come up with new ideas that people reject first because they haven't done it before (that's why blind faith). But it's also important to take a step back and make sure you have a viable plan that resonates with your customers (avoid false hopes).
We are fortunate to have builders who challenge each other to cycle through customer feedback. Product development often starts with the customer: it must be released to the media in a timely manner (to protect the customer's interests) and a FAQ document (to detail how we will build the project) to help dispel the blind belief in false hopes (at least in general).
5. Define a "minimally cute product" (MLP) and iterate quickly: Figuring out where to draw the boundaries for a release is one of the most difficult decisions the team has to make.
Often, waiting too long, or sticking with too many fancy features, makes the team miss out on the first-mover advantage. In addition, in a fast-growing track, it is necessary to establish a reputation before the best peers are far ahead.
A released product has to be good enough that you believe it's going to be loved from the start (which is why it's called "the least lovable product" rather than "the least viable product").
But on newer tracks, teams often do a better job of delivering this MLP to customers and iterating quickly thereafter.
6. Take a long-term positioning: We are sometimes criticized at Amazon for not closing many businesses. This is because we are more tolerant of investment than most companies. However, everyone knows that revolutionary inventions take many years.
If you've made a big bet and believe it will significantly improve the customer experience (and your company), you'll have to be involved for the long term, or you'll usually give up quickly.
7. Be prepared to fail: If you invent a lot, you will definitely fail more than you would like. Nobody likes this part, but it's the way to go.
When we're rolling out something that clearly doesn't work, we want to make sure we learn from what doesn't go well and provide a great foothold for team members who are doing well, or your best employees will be hesitant about new initiatives.
Albert Einstein is considered a being who "describes compound interest as the eighth wonder of the world" (those who understand compound interest can get compound interest, and those who do not understand it will pay a price).
And we think iteratively innovating in the same way to create miracles for our customers. Constantly inventing and improving products for customers has a "compound interest" effect on customer experience, and reverses the development prospects of enterprises.
When you start the compound interest effect, time is your friend. Amazon is a big company with a lot of big businesses, but for us, the company is still in its early stages.
We will continue to be innovative – innovating in the industries in which we operate, innovating in new industries that have not yet been launched, innovating in new ideas that we have not even thought of, "Always Day 1".
