Hacking Together Prime Video

3 product hacks, outnumbered, Jeff Bezos calling it “The worst product name in Amazon’s history", press leaks, customer complaints, and legal problems—all part of Prime Video's launch (Chapter 2)

Hello, it’s Ethan & Jason. Welcome to Level Up: Your source for executive insights, high performance habits, and specific career growth actions.

Three FYIs:

1. Watch Ethan’s interview with Aakash Gupta (Product Growth). The Magic Loop framework to get promoted, leverage, and Ethan’s business. Watch here.

2. Get Jason’s networking question bank. A combination of career focused and casual ice breaker questions. Get it here.

3. Read (archived post) Relentless Self-Improvement: 4 ways to be exceptional.

---

Chapter 2: Hacking Together Prime Video

The original version of Prime Video was built by a scrappy, under-resourced team in 17 months. It was both a great accomplishment and an incredible disaster.

This is Chapter 2 of the “Story of Prime Video,” and in it I will explain how we built the first version of Prime Video despite a number of significant hurdles. If you haven’t read the first part of the story, where I talk about the project’s inception, read last week’s newsletter here.

To recap, the hurdles my team of 9 was facing when setting out to build Prime Video were:

1. No dedicated Windows machines to build our PC application.

2. No dedicated functionality in the product catalog system.

3. No dedicated time from the one UX designer supporting all of digital media.

To solve these problems, we devised a series of non-traditional solutions that allowed us to avoid long waits for traditional resources and to continue building quickly.

I call these solutions the “product hacks.”

Product Hack 1: Run Windows in the corporate IT data center, not the standard Linux production data centers that Amazon used.

As I mentioned in the first part of this story, Amazon was almost a pure Linux shop. However, our online video service needed a PC-based application, and we needed Windows servers to build it. The Linux-based data center teams did not want to run a new OS, but we were able to convince the corporate IT team, which ran Windows machines for Outlook email service, to host the machines for our team.

This was an out-of-the-box solution and a good example of influence; I wasn’t able to “order” the corporate IT team to let us use their machines, but I was able to convince them it was in the best interest of the business.

So, the first product “hack” was comprised of two parts:

1. Identifying a creative, non-traditional solution.

2. Influencing cross-functional teams to help us implement that solution.

Product Hack 2: Representing digital media as clothing.

Behind each item you can buy from Amazon is a database called the catalog, and even in 2005, this catalog contained millions of items with many different attributes.

The catalog was a fragile legacy system from Amazon’s earliest days, and there was a huge number of pending updates that were going to take precedence over our video product.

Basically, we were going to have to wait a year to get the proper functionality for our product. Since adding a year to our schedule wasn’t an option, we started looking for alternative methods.

This is what we came up with:

TV shows are normally represented as a hierarchy. The show itself, like Game of Thrones, is the top level, and within this you have seasons of the show. Seasons are composed of episodes.

So, while movies could go straight into the catalog, we needed something in the catalog that could represent the hierarchy of a TV series. .

We found our solution with clothing. A particular dress or shirt was the top level, like a TV show. Under that, clothing could come in multiple colors, so seasons could equate to blue, red, green, etc. Clothes also come in sizes, so each episode could be represented by a size.

So, while consumers saw a TV show, seasons, and episodes, these digital products were represented on the backend by articles of clothing, colors and size. Each TV episode had a dress color and size behind the scenes.

With this “hack,” we were able to get our digital products onto the catalog without waiting a year for a custom build.

Product Hack 3: Engineer-driven UX.

As I mentioned in the first chapter, the third significant hurdle was that the whole digital media group had only one UX designer supporting three separate businesses.

And, out of the three, the video business was the lowest priority. Much like the catalog build, we were going to have to wait a while before the designer could help our team directly.

So, we executed what was the most “behaviorally informed” hack of them all: We built the UX ourselves. Not being designers, it was not up to Chip’s professional standards. So, rather than let us launch with an inferior design, he helped us to bring it up to a professional level.

Nothing will scare a designer more than letting a bunch of engineers taking creative control!

With those three product hacks, we were able to clear the three technical and organizational hurdles. We had windows machines, we could get on the catalog, and we had a user interface.

But, even with those hurdles cleared, there was a significant motivational hurdle that took the wind out of our sails during the build…

Outnumbered, outmanned

During our product cycle, a news article came out in Wired about how Yahoo had a team of 400 working on an online video product. When our small team read that article, we felt pretty hopeless. How could we compete with such an army?

Of course, it turns out that Yahoo never ended up shipping anything of significance in online videos. The leader of the effort had a falling out with the Yahoo founders, and the project imploded. While Yahoo is the most significant example of such failure, there were several other competing efforts that also fell by the wayside. By contrast, small startups such as YouTube, which was tiny at the time, became threats, particularly after their acquisition by Google.

Lesson: Big teams and big spending do not guarantee success. They sound good and are very tempting, but they come with their own problems. It’s better to find product/market fit, then scale.

This is exactly what made smaller teams like ours (and YouTube’s) more successful.

Progress and Problems

With all the hacks in place and the threat of big competition looming large, our product made rapid progress. By the summer of 2006, about 16 months after I joined, we were approaching release. We had overtaken the music project, which had decided at the last minute to change business models, and the Kindle project, which required hardware development. Along the way, we had added three engineers.

The next major product decision to make was a name.

Amazon had little experience in creatively naming products. For example, the internal name of the division that sold Books, Music, VHS tapes and DVDs was “BMVD.” Fine for internal use, but not exactly the kind of name that makes a big splash in a public release. So, we hired a branding agency to help come up with possible product names. While I do not recall all the options, I do recall three we talked about.

1. One was “Zeezeebio” or “Zezebio.” The approach here was to use a nonsense word that we could own and easily trademark. We could then build recognition around it, like “iPod” or “Coke.”

2. The next, signifying that digitally delivered media was a new way to watch a movie, was “Rebirth Media.” Our product would be the rebirth of watching movies in a new way. This name died when someone called it “Afterbirth Media.”

3. The third name was Unbox, or Unboxed. The idea was that the video would come without a box, making it different than VHS tapes and DVDs.

After some debate, we settled on Unbox Video.

Jeff Bezos was in the room and agreed with the final choice.

He would later call this “The worst product name in Amazon’s history.”

Unbox Video

As we approached launch, I went to my VP to plan a beta cycle. With 15 years of experience, I was accustomed to running beta tests and knew that in-house QA could never find everything.

But, my VP told me that there would be no beta testing. The company wanted to keep it’s entry into the digital media space secret for as long as possible, and therefore wanted our launch to be a surprise. I protested this decision and was told that I could recruit internal Amazonians to test the product, but only on their corporate laptops.