- Network Effects (a16z) and Data Network Effects (Matt Turck)
- Full stack Startups (Chris Dixon), Full Stack Startup Index (Anshu Sharma)
- Stack Fallacy: Why Big Companies Keep Failing (Anshu Sharma)
- Disruption’s Long, Slow, Complex Journey (Steve Sinofsky)
- Disruption is not a strategy (Jerry Neumann)
What is the ecosystem effect?
- Social network effects (FB, Snapchat)
- Data network effects (Uber, Palantir)
- Ecosystem network effects (Google, Uber, Netflix): This is a higher-level version of the full-stack startup (defined by Chris Dixon), if you will. This is how you create data network effects on a platform itself. Original thought leading to conversation: Companies that rebuild an industry, rethink the experience, collect a bunch of data, then use this in product decision to improve their offerings and beat incumbents.
Examples of Companies Exhibiting “the Ecosystem Effect”
- Initially built the search engine, on top of previous innovations in computing (mainframe, OS, internet browsing).
- Built on top of the search engine with AdSense, ad tools to help grow and scale their ad business and revenue
- Expanded horizontally with Gmail, YouTube, Drive, Music. In the process, built OAuth and single-login to all apps as well as slick integrations between products (ie easily embed Drive docs in Gmail).
- Eventually went back and rebuilt layers of stack below them (rebuilt the browser - chrome, rebuilt the OS - ChromeOS, rebuilt the PC - Chromebook, the internet - Fiber.
- Building below and horizontally are usually moonshot projects, that with traction, become full-blown projects (ie Gmail, Trends, Adsense)
- Google is going into other industries and verticalizing their new products
- "No One Ever Got Fired for Buying Google" is the new "No One Ever Got Fired for Buying IBM" (credit to Zach Hamed for the comparison)
- Originally started as a DVD rental company through the mail. After seeing the potential of streaming, started investing in scalable cloud technology to efficiently deliver video over the web.
- Now that some previous content licenses are expiring, Netflix is investing in buying shows, some new, some old, and offering them as original content (one level down in stack).
- Eventually, Netflix will run their own studio (another level down in stack) and produce a good majority of their content.
- Most recent example (company is getting to the stage where it’s becoming LARGE)
- They started off with a simple goal - fastest way to take you from A to B. Expanded product offerings vertically with uberPOOL and uberEATS.
- Now are integrating up to autonomous vehicles in Uber, the step to eliminating human labor from their system. After Otto acquisition, it seems as if they’re integrating on that horizontally (autonomous uber, but for commercial trucking)
- As per Semil Shah,
it can apply those resources [from Uber China - Didi deal] to technologies “up the stack” for a world in which your Ubers are autonomous — that could be pods or cars, sensors, robotics, mapping technologies, deep learning, and a host of other requirements to make a fully-integrated self-driving network a reality. With 80% of each fare you pay going to your driver, the company has a huge incentive to bite into that for its next big meal. 
- Why did they buy Quip? Why did they buy Heroku? Seems unrelated from sales software, but they’re building an ecosystem of products.
- Apple seems to be creating an ecosystem of brands. With the recent Beats acquisition, Didi investment, and potential McLaren purchase, they want to create a collection of brands with potential collaboration between them. Apple might power the software in the McLaren, that is exclusively on Didi, and has Beats-branded audio in-car.
Case Study: Dropbox vs. Google Drive
You could take this from a physics perspective and think of the stack as a hill. As you get to the top of a hill, potential energy increases, and kinetic energy starts to decrease. After you pass the monument of building lower in the stack (at the bottom of the hill), you’ve built up enough potential energy to build horizontally from there, since the previous infrastructure already exists. When Uber created UberEATS, a lot of work was already done, since a network of drivers already existed on the road (rather than having to build from scratch).
Wiley and I recently recorded a podcast on this phenomena where we explain the idea in more detail + clarity: