Amazon's Success: lean thinking and practice as drivers

Contrary to what some people think, adopting the lean system in a high-tech environment is not just possible; it is essential. Marc Onetto, Amazon's former senior vice president of operations and customer service, pioneered the application of lean management to retail distribution at the company. In this article, he shares the many ways in which lean thinking and practices helped improve Amazon's operations.

As with any company that adopts the lean mindset, Amazon's driving force was its constant and unwavering dedication to its purpose: customer focus. From that foundation, the company applied lean to identify and eliminate muda for the customer – from how it built its website to how it fulfilled orders and much more.

When asked what the lessons learned from lean at Amazon were, Onetto said they learned something very important along the journey (to hear Onetto talk about this journey – in English –, click here): lean and high technology can work in synergy to eliminate muda. From the start, many of the sensei in Japan had serious concerns that information technology could add waste instead of eliminating it, simply speeding up bad processes rather than improving them. What kept Amazon from falling into that trap was the fact that, in addition to good technology, they already had a very strong customer focus. This allowed them to truly combine lean with high technology. Another thing Onetto mentioned when talking about his learnings was that key lean concepts, such as andon, can be extended to different businesses, as Amazon did by stopping the line in the customer service center. See below what Onetto has to say about the company's lean journey.

Customer service with value stream mapping

After introducing lean in the fulfillment center, we extended it to the customer service centers. One interesting thing we did was map the value stream. By doing so, we noticed some cases that caught our attention: when receiving a call from a customer, the call center agent had to look up information in the system. We counted the steps he had to take and found that sometimes, for example, he would press the “Enter” key as many as 29 times before he could find the information needed to answer the customer's question. So, of course, there was a lot of waste there. Therefore, we rethought all the service systems from the customer's perspective. What kinds of questions usually came in? That's how we tackled this waste of excessive clicks the first time we did a kaizen, reducing that number to 8 clicks.

The management of the future starts at Lean Summit! Join the 2025 event and explore new frontiers. Register now!

Defining customer-centric principles

In fact, we wanted to define what we called customer-centric principles for our call center, and the main goal was to treat customers like friends or even like family. Some people worried about dishonest customers trying to take advantage of this, but our computer system is usually able to prevent that. It's true that there are people who try to steal money from Amazon, but even so it's worth taking the risk to be honest with the customer, because that's where you gain a lot (actually, not many people try, not only in Japan, which is the most honest country in the world, but also in other countries, such as France, Italy or the United States; at the end of the day, the percentage of people who try to steal from you is not that large).

There's a story that happened one day that I like to call The New York Times customer-centric Christmas miracle: a person called one of our agents, who was in the state of Washington, and said: “I have a problem: the website says my package was delivered, but I think someone stole it from my porch. I'm divorced, and my kids will be here tomorrow night. It was a present for them”. The agent said: “No problem, sir. We'll take care of it. We're sending you the same package right away. You're in New York, so we'll ship it straight from the fulfillment center, and it will be there tomorrow afternoon. You'll have your package for your kids”.

He then said: “Thank you. How much do I owe you?”. And the agent said: “You don't owe us anything”. He protested: “But it wasn't your mistake; the package was stolen”. The agent reassured him once again: “That's fine. To us, you're a friend. We like our customers”. What the agent didn't know was that this “customer” was a journalist from The New York Times. So, that day, we got a positive article in the newspaper. That's what I call The New York Times customer-centric Christmas miracle.

Applying andon to customer service

Then we took it a step further. We adopted the andon concept. As you may know, in the Toyota Production System, it is essential to empower line workers so they can stop the line when they detect a defect. Anyone who has had the chance to be trained by a Toyota sensei has probably visited a plant and seen 5,000 people stop working because one worker, assembling something somewhere, such as a car's wheel or headlights, stopped the line by pulling the andon.

We brought this concept to customer service, and we gave agents the authority to stop the line. What does that mean? In our case, it means that we detect, using a computer, using technology, when a customer is calling about the same defect that has already occurred with another customer. So this is becoming a bigger problem. We alert the agent that it may be a defect on our side. Then the agent is empowered to stop the line, that is, he can press a button on the computer, and the item will show as unavailable on the website, sending it to the quality lab. The agent doesn't need to ask anyone for permission to do this.

Solving problems at the gemba

Of course, we have a process in place to restart the line. We do a weekly andon review with retail to look at items that have stopped the line for more than two weeks, to understand what is happening and why we haven't been able to solve it. One example I remember now was a SKU error: a customer ordered a headset but received a hard drive, which is a little hard to put on your ear. It was the second time this had happened, so we stopped the line – the agent stopped the line –, and we went to check. We pulled all those items from the fulfillment center and realized there was a SKU problem. The same factory was producing hard drives and headsets, and they had put the headset SKU on the hard drive. So the associate, when scanning, didn't know he was shipping the wrong product.

This andon process, which still exists at Amazon today, has eliminated hundreds of thousands of negative customer experiences. And [Jeff] Bezos loved it, which makes a big difference, because when I stopped the line for the first time, I remember thinking: “I think they're going to fire me for doing this, because we won't produce anything else until we find the defects”. The truth is that Jeff loved it. At first, I had to ask: “Give me a little time to put the process in place”, but he loved the fact that we would detect customer defects very quickly through this process.