Working Backwards: Insights, Stories, and Secrets from Inside Amazon

by Colin Bryar

At Amazon, understanding what’s normal is the responsibility of the metrics owner, whether that’s an individual contributor or a manager of thousands. Many statistical methods, such as XMR control charts,3 can highlight when a process is out of control. For us, however, experience and a deep understanding of the customer most often turned out to be the best way to filter out the signal from the background noise. For the most part, metrics are reviewed daily by their owners and weekly in the WBR, so that expected fluctuations become familiar and exceptions stand out.

My team and I quickly learned that invention is a more challenging path than fast following. The roadmap for fast following is relatively clear—you study what your competitor has built and copy it. There is no roadmap for invention.

When Jeff asked Steve to run Digital, he also changed the org structure at the top of the company. Previously, Steve had reported to Diego Piacentini, SVP of Worldwide Retail, who reported to Jeff. Now Steve reported directly to Jeff, a clear sign that Digital was a high priority. There were two important benefits to this approach. First, this meant that Steve was not encumbered with the many responsibilities that went with managing any of Amazon’s then-current businesses or operations; he was given the autonomy and authority to devote single-threaded focus to Digital. Second, it meant that Diego and his peers would not be required to spend any of their time on Digital. They were free to continue to devote their single-threaded focus to building our retail and marketplace businesses as well as our fulfillment network.

On the organizational side, we used the two-pizza team structure, which allowed our Digital teams to not be dependent on or a distraction to the engineering and business teams running the retail and marketplace business. Our people were autonomous with respect to their ability to achieve the goals that they had agreed to with Jeff. From Jeff’s point of view, this meant he wouldn’t be stymied by arbitrating resource conflicts and dependencies at the ground level. He could hold each two-pizza team leader accountable for staffing their team and achieving their goals.

We applied the new two-pizza structure to every part of the org chart below Steve and his direct reports. The two-pizza structure became more complicated at the top of the org chart.

We decided that there would be separate leaders for business and tech for each digital product category—books, music, and video. Each of these category leaders would hire leaders for each business function, such as product management, marketing/merchandising, and vendor/content management (licensing digital content from publishers, studios, and record companies). Each general manager (GM) category leader had a corresponding peer leader on the engineering side. Each engineering category had a two-pizza team for each major component of the software services (e.g., content ingestion and transformation) and for client application software. This was mostly a pragmatic decision based on the skills of the leaders. For example, I had no experience at that time managing an engineering organization.

In my case, this meant that by 2005, instead of leading the business team for digital books, music, and video, I was focused on leading just music and video. In 2007, my scope grew when I took on leadership of the engineering organizations in addition to business. This process was continuous as each year changes were made where the scope of the work had become too broad to break up or divide teams into subteams.

in 2004, video client application development was handled by one two-pizza team. It then became three teams, one for web, one for mobile devices, one for TV devices. Then the mobile team became four teams (iPhone, Android phone, iPad, Android tablet) and the TV team became five-plus teams (Xbox, PlayStation, TiVo, Sony Bravia, Samsung, etc.),

The radical part was that these teams were established outside of the then-current retail and marketplace business and engineering organizations, and we were thinking big and long term by hiring and building a large organization to support three speculative, new businesses.

So, although we knew nothing about building hardware, Jeff and Steve decided that the place to start was at the consumption end of the chain: hardware, specifically e-books. There were multiple reasons for this. One was that books were still the single largest category at Amazon and the one most associated with the company. Music was the first category to move to digital in the marketplace, but Apple had a big head start and our sessions did not produce a PR/FAQ for a music device or service idea that was sufficiently compelling. Video had not gone digital yet, which seemed like an opportunity, but it became apparent that there were a number of barriers to creating a great video experience at that time. These included getting the rights from the studios to offer their movies and TV shows digitally, the time it would take to download massive video files over the slow (at the time) internet, and uncertainty about how consumers would play these video files on their TVs.

The idea that Amazon, a pure e-commerce distributor of retail products made by others, would become a hardware company and make and sell its own reader device was controversial. Like nearly everyone at Amazon outside our Digital leadership team, I had a really hard time accepting that it was a good idea to make our own hardware when it seemed that nearly every leader in the company and on the board of directors was questioning it.

The second point he made was that if you decide that the long-term success and survival of your company, like any company at a crossroads, is predicated on having a specific capability that you do not currently have, then the company must have a plan to build or buy it.

In 2004, the U.S. retail industry was estimated to generate more than $3.6 trillion of sales, of which less than 2 percent was conducted online. Amazon’s growth rate was slowing, but the shift from offline to online commerce was accelerating. That meant one thing: if Amazon’s growth continued to decelerate, the company would become a smaller and smaller player in online commerce over time. We were determined to find a way to reverse this trend. What would it take to get Amazon’s top-line growth back on track?

We conducted surveys of new customers, existing customers, people who had shopped online but had not yet done so on Amazon, and people who had never shopped online at all. We asked them for the top reasons why they didn’t place an online order, and what would make them shop online more frequently. In every survey, the top answers remained the same: one of the biggest reasons people didn’t order online was that they didn’t want to pay for shipping.

So Super Saver Shipping reduced our costs and lowered prices for the customer. The program, which seems laughably primitive today, offered a valuable insight into what our customers wanted. They were delighted to be given a free shipping option, even if the trade-off was “slow and free” or “fast and expensive.”

Now that customers had gotten a taste of free shipping, they no longer wanted to be forced to choose between “slow and free” and “fast and expensive.” Jeff exhibits discomfort when presented with an either/or proposition in which both results are mediocre. Viewed through the Customer Obsession and Insist on the Highest Standards leadership principles, the only answer to the question, “Which would you rather have, ‘slow and free’ or ‘fast and expensive’?” is “fast and free.” So the catch was that “fast and free” was where Amazon needed to go next, but our fulfillment capabilities were not up to the task.

Crafting an affordable program that we felt would lead customers to buy more—rather than continuing on with our tried-and-true method of funneling cost savings back into lower prices—seemed difficult, if not impossible, at the time, especially given the constraints of our fulfillment and delivery supply chain.

Once a plane takes off, its empty seats have no value. Therefore, airlines, in exchange for loyalty, can give away marginal inventory that would otherwise go unsold. Whereas in retail, giving away either product or shipping fees always has a cost. None of the ideas made it very far because they could not meet the three essential criteria.

We worried that the membership programs that involved an annual fee for free standard shipping or for free two-day shipping would encourage customers to purchase fewer items per order. This would not generate enough money to cover shipping costs, which meant it would not be sustainable. Unsustainable growth at any cost was definitely not the customer behavior we were after. Review meetings often ended when someone asked, “Wouldn’t the money be better spent on lowering prices and improving In Stock for customers?”

Another concern that caused a big debate was how heavy buyers would respond to a loyalty program. Would it encourage them to place additional orders? Or would they have placed the same orders even if they had to pay for shipping? The purpose of this program was to drive incremental buying behavior, not for Amazon to pick up the shipping tab as a way of saying thanks to our big customers.

The institutional no refers to the tendency for well-meaning people within large organizations to say no to new ideas. The errors caused by the institutional no are typically errors of omission, that is, something a company doesn’t do versus something it does. Staying the current course offers managers comfort and certainty—even if the price of that short-term certainty is instability and value destruction later on.

It’s what causes a board of directors to say no to a big change of strategy (think Nokia and Microsoft missing the turn on smartphones). It’s what drives frontline managers to keep their top performers working on a current project and say no to their involvement in high-risk experiments that could fail but could also pay off handsomely later—

Fast Track had been developed to allow the fulfillment system to make more precise estimates about shipping time—that is, we went from “Usually ships within 24 hours” to “This item will ship tonight if you order within 1 hour and 32 minutes.

The first customers who joined were the heavy buyers who already spent more than $79/year on expedited shipping. So we were really just subsidizing their existing habits. Though we had created a game-changing online shopping experience, shifting consumer behavior takes time.

So when Netflix launched Watch Now, we took note and discussed the service in detail, but the prevailing wisdom inside our team and among others in the industry was that it was a half-baked test, not a serious offering.

Jeff Bewkes famously said that the Netflix threat to Warner Bros., where he was chairman at the time, was about the same as the threat of the Albanian army to the U.S. armed forces.

We had been selling these same TV shows on Unbox for $2.99 an episode, which was cheaper than buying a DVD, and making them available the day after broadcast. Now, with Hulu, you could watch many of the same shows, just as soon, and for free.

We had no ability to influence the studios on this matter because two of them—News Corp and NBC Universal—owned Hulu.