Amazon Cloud CEO Adam Selipsky has been there almost from the beginning: He joined in 2005, left in 2016 after 11 years at Amazon to run Tableau, and became Amazon Cloud CEO in 2021. Andy Jassy, the former chief executive of Amazon Cloud Technologies at the time, succeeded Jeff Bezos as Amazon's chief executive.
The following is the content of an exclusive interview with Amazon Cloud Technology CEO Adam Selipsky by the editor-in-chief of the technology media The verge.
How to Make Decisions in Amazon Cloud Technology
Reporter: Amazon does have a very strong set of leadership principles and a very clear decision-making process. It sounds like you're saying that you want to be in a "Day one" mentality at all times, as opposed to a "Day two protection mentality. What is your decision-making framework? How do you make decisions?
Adam: This is a hard question to answer in the abstract, but let me try. You mentioned Amazon's leadership principles, and we have 16 of them. I sometimes call it Amazon's operating system. We apply these leadership principles when recruiting. So if I were doing a round robin interview, I might be assigned to "hold the highest standards" or "think big" or "learn and be curious." I would actually be interviewing for this leadership principle. It really became part of everyday speech, part of Amazon's vocabulary. They are very important. If I had to choose one, Amazon’s true core is its customer-first leadership principle.
To answer your question, I think the way we (myself of course) make decisions is always from the customer's point of view. By the way, there are misconceptions - I've learned that when people say customer-centric, or we say customer-obsessed, they mean something different. I think a lot of people think the way you show that is on an emotional level. "Don't I like my customers?" As far as I know, there are some very traditional IT competitors who don't seem to like their customers. Or you might like your customers, or you might love your customers, which people think is true customer centricity.
But what I learned is that you can’t measure this on an emotional scale, and that the most customer-centric thing you can really do is twofold. First, get to know your customers deeply in a way that most companies don’t take the time to do. Sure, they send out a survey report or have product managers talk to some customers, but they don't understand exactly what problems their customers are having or what exactly they think about the product you've built so far. view.
And then the second part, which is actually harder than the first part, is taking that understanding and really putting it at the center of the decision. It's that simple. You have a complete understanding of the customer, and then when you price something, you say, "Oh, well, what's the best way for me to maximize profit? What do I need to do?" I see that in a lot of cases Companies, when they make their most important decisions, they just routinely leave customer information at the door.
When making decisions, we make sure to keep the customer perspective at the center of the room. The way we do this is by working backwards. So any time we were building something, not just a big new service or even a medium-sized feature of an existing service, we would actually write a press release before the developers started coding. If we can't describe in simple words what's exciting and groundbreaking about this thing we want to build for our customers, then why on earth are we wasting our time building it?
You can also clear up all kinds of misunderstandings between the teams. Are we building this for developers? Are we building it for IT staff? Is this for line of business users? How high up do we build in the stack? Will it build on our original service? Do we need new technology? Here's a press release, followed by a detailed FAQ behind the press release. We do this dozens or hundreds of times a year. That way we know we're about to start building something that will at least deliver great performance to our customers. This may be at the heart of our decision-making.
Another thing for me personally is to hear a lot of voices. I really enjoy taking in a lot of different perspectives. I don't think the people closest to me, the most senior, always have the best ideas, or the sharpest opinions on something. I would push and ask people to justify and defend what they said. That's very important to me and helps us get on the same page. Or, if we can't agree, at least empower senior decision makers with as much knowledge as possible to make a decision.
How to Make Decisions in Amazon Cloud Technology
Reporter: Please tell me how you brought Amazon cloud technology to a new scale? We want to reduce the size of the organization. Is this your decision? How did you put it into practice?
Adam: First of all, I will say that any time you have to lay off people, it's going to be very painful. You're dealing with people's lives, their livelihoods, but also their families. We take it very seriously and understand the impact it has on people. So I don't want to minimize that in any way. The number of employees at Amazon Cloud Technology is growing extremely rapidly. Just from 2020 to the end of 2022, Amazon Cloud Technology added tens of thousands of employees. And then, earlier this year, as we thought about the overall economic uncertainty, the macroeconomic environment and we really wanted to focus on our most important priorities, we did end up making small, single-digit layoffs.
We made the decision: How can we become more efficient while being confident that we still have a lot of potential for innovation? We try to become increasingly clear about what our real priorities are. We grew so fast, so we did a lot of things. I think any time you're in a situation like this, every once in a while, step back and say, "What are our top priorities? What are the most important services to our customers?"
In many cases, we just moved people and teams around so that we could focus on our most important priorities this year—not because some of the things were bad or because they were bad ideas, but just because we decided to focus on the most important ones. important things. In a few cases, we will eliminate these roles if we do not have sufficient skills required in a specific area. In our highest priority areas, we still have vacancies we are hiring for, and the people we are making redundant do not have the skills that are available.
Reporter: You are talking about reorganizing Amazon cloud technology and redoing priorities. What is the current architecture of Amazon cloud technology?
Adam: We have been and continue to be very decentralized in terms of our product. We optimize for innovation and speed. Speed is often underestimated. I think people greatly underestimate the importance and power of speed. They're also very fatalistic about it. I often hear clients say, "We're just not a fast company. We don't have the ability to be agile." I tell them that speed is a choice. You choose how quickly you move, and that choice takes a lot of factors into consideration: how you organize, how many and what types of people you have, and what senior leaders insist on from their teams.
There is a series of things that will happen soon. One of them is organization. We chose to build what we commonly call separable teams. So we want the team to be as independent as possible. Of course, teams do depend on each other now, but it's also possible to have fewer dependencies. We chose to break down and restructure the teams as much as possible to make them as autonomous as possible. They control their own destiny as much as possible.
Another key concept is single threading. If you take an existing successful business and the leader of that business and then give that person a new project to work on, it's almost inevitable that the project will struggle because they have the revenue streams, business, and operations to stay afloat.
What we tend to do instead is take super-successful leaders out of what they’re doing and let them focus on something new. That way, it gets 100% of its attention. So we do have a lot of small business or product area general managers who are responsible for both development and product management, those types of functions. When unified under one leader, they can move much faster than if we had some of the large, single-function structures.
Now, in terms of going to market, we don't want to show up to customers with 200 separate services. So on the go-to-market side, we're structured more around organizations by industry vertical or geography. But we always have an account owner who acts as a facilitator for the client, bringing in various experts for different products or different types of technical topics. This way we can show as much of Amazon's cloud technology as possible.