Dissecting Amazon’s Platform Play

Question: Is Amazon a Pipe or a Platform or both?

Short Answer: It’s complicated!

Long answer follows…

Pipes and Platforms are two contrasting business models, as we noted in the last post. However, the internet itself is a platform on which others (you, me, web developers, app makers, everyone) create value. By virtue of this fact alone, every business on the internet exhibits some Platform characteristics, even if it may appear to be a Pipe. Platform Thinking is an approach to creating internet businesses with platform characteristics.

The question above was raised by one of the readers as a comment to the previous post and I felt it needed a post in itself as an answer.

 

Let’s start with a framework in mind. There are three broad properties of a platform:

Magnet: A platform needs to get both producers and consumers on board

Toolbox: A platform needs to provide the tools required for producers and consumers to interact (and transact)

Matchmaker: A platform needs to match producers and consumers, leveraging data

 

Let’s take a quick look at Amazon with this lens. I will deliberately be taking an over-simplified view of some of Amazon’s business lines here as this is meant to be purely illustrative.

 

Amazon, the Pipe…

Amazon started out as a Pipe. Establishing the online store model, it managed the producer role itself by sourcing products, managing inventory and selling them down the pipe, Amazon.com. A very linear model, it simply brought the offline store model online and users continued to be customers, except for one small difference.

Verdict: Store = Pipe

 

… with elements of a platform…

Amazon did set out building its main business (of selling books etc.) on a pipe model. However, Amazon first dabbled with a platform model on its reviews ‘feature’. Reviews could be created by users and consumed by other users, and in that sense, Amazon allowed creation of value by producers (review creators) on a platform (Amazon.com). The main business continued to be a pipe but it showed platform characteristics with its reviews. A business using Pipe Thinking would just have sourced expert reviews but Amazon used Platform Thinking to create an entirely new source of product reviews.

Verdict: Reviews = Magnet (+ Toolbox)

 

… and data-driven efficiencies…

Unlike pipes, platforms are intelligent. Also, platforms exhibit network effects of data. The more the number of users using a system, the more valuable the system becomes for every individual user because of the usage data that it collects.

The second way in which Amazon used Platform Thinking was to enable the “Users who liked this also liked this…” feature. This was an instance of the Matchmaker role where Amazon used data about consumer usage to match them with the right products. This collaborative filtering model became more accurate and, hence useful, as more users used it.

Network effects of data are absolutely non-existent in traditional pipes in the offline world. A TV program doesn’t get more interesting as more people watch it. (In today’s world it does, on a second screen: a mobile app, again working on Platform Thinking).

Verdict: Collaborative Filtering = ~Matchmaker

 

…transitions from a pipe structure to a platform structure…

Amazon’s main business eventually moved away from a Pipe model to a Platform model when Amazon launched Amazon Marketplace allowing external merchants to sell their goods via Amazon. Amazon continued to be a producer, but allowed other producers to also create and transact on the platform. This was the key transformation that moved Amazon’s business model completely to a Platform model.

Verdict: Amazon Marketplace = Magnet + Toolbox + Matchmaker

 

 

…and allows extenders to extend the platform…

Another hallmark of Platform Thinking is the fact that external parties can extend the properties of the Platform. Referencing the three-part framework mentioned above, Amazon extended its platform in the following ways:

Affiliate Program and Widgets: This allowed users to play the Magnet role and get other users to Amazon, by peppering links to Amazon across the internet. Given that users played the Magnet role, Amazon compensated them as well.

API: API developers could extend the functionality of the platform.

Verdict: Amazon = Extend(Magnet) + Extend(Toolbox)

 

…creates the device-as-a-platform…

Finally, with the launch of Kindle publishing and the Kindle hardware, Amazon applied Platform Thinking to the publishing world. The publishing world has long existed on a Pipe model where editors control access to the Pipe. With Kindle Direct Publishing, Amazon has created the largest publishing platform allowing authors (producers) to directly publish and access a market of readers (consumers).

Verdict: KDP = Magnet + Toolbox + Matchmaker

 

…and allows extenders to extend the device-as-a-platform…

Kindle started as a reading device but, with the launch of Kindle fire, has been reinvented as a full-featured tablet. Amazon now allows app developers (platform extenders) to create applications specifically for the Kindle Fire App Store.

Verdict: App Store = Extend(Magnet) + Extend(Toolbox) + Extend(Matchmaker)

 

… and decides on extending its infrastructure as a toolbox for all…  

Finally, with AWS, Amazon realized that the infrastructure that it had set up to power its own operations, could be offered on a subscription model to developers externally. A Toolbox of sorts, AWS provides the underlying tools and technology for developers to create their respective applications on top of it.

Verdict: AWS = ~Toolbox 

 

 

Takeaways

Amazon is a great case study to illustrate several facts:

1. Even while acting as a Pipe, an internet business can show Platform properties (e.g. user-generated reviews).

2. Any system that uses data can leverage network effects and become more useful as more users use it. Traditionally, we associate network effects only with full-fledged Platform models. However, Amazon’s collaborative filtering feature shows that network effects can exist even on a linear model if it uses aggregate data to create value.

3. Pipes often transition to Platforms. Amazon opening up, first its marketplace, and then its publishing platform, demonstrates a great way of building a platform without having to worry about the chicken and egg problem. Start as a Pipe, get consumers hooked, and use that traction to attract producers and open up the Platform.

4. Many internet businesses may exhibit only one or two of the three platform properties. AWS only acts as a Toolbox. TripAdvisor (an independent business similar to Amazon’s review feature) plays more of a Magnet role only.

 

About the Author:

×