The “data network effect” is a term used to describe the relationship between data, product performance, utility and user retention. While this may sound a bit complicated at first glance, it gets more intuitive as we look further.
We start off with the number of users of a specific product or service — a social media platform, for example. The more users who sign up for this platform, the more data is fed into the system.
Each individual user contributes a wealth of personal information, while the body of users as a whole represents a huge trove of data. With this increase in data, the product or service becomes smarter, learning more about its users, how they interact with the service, and can better understand how best to serve them and prioritize certain features over others.
It does this through the use of algorithms that comb through data; the more information available, the more effective the algorithm. With smarter algorithms comes a better overall product, which in turn attracts more users, and the cycle repeats itself.
It is easy to see how this can quickly bloom into something much larger than it began as, ultimately becoming a service that is so indispensable to the lives of its users that it is sometimes difficult to imagine life without it (consider Facebook, as an example).
The data network effect is not without its issues, however. All data is valuable, yet some data is more valuable than others. There comes a point in the acquisition of data that it does not always mean smarter algorithms and, therefore, a better product — a “peak” data point. For the data network effect to continue its cycle, the algorithms need to continue to improve as well, putting much of the onus of operation on the developers and technicians who write and maintain them. For Google, it is not the data that we give it that matters most; it is what Google does with the data.
Enter AI, where algorithms can maintain and improve themselves in order to keep the data network effect moving. As algorithms become self-teaching, an individual service may not improve with more data, but better algorithms may inspire new services. Data may peak in terms of one product, then create an entirely new product that drives its own data network effect.
In the real world, companies use the data network effect to inspire acquisitions of other companies or services and therefore drive an influx of new users. When Facebook acquired WhatsApp in 2015 for a hefty $19 billion, it was not doing so just because it wanted to add a messaging service to its social media offerings; rather, its primary interest was more likely the massive incorporation of new users and the data they brought. This provided Facebook with new insight and algorithms that understood how users interacted with each other and what was important to them, ultimately improving Facebook itself.
Hypothetical future acquisition also show the appeal of the data network effect. The proposed merger between Groupon and Yelp, for example, would combine two huge and complementary user groups: Groupon’s shoppers, service seekers and other deal hunters with Yelp’s reviewers and advertisers.
This could foreseeably create an environment where customers interact more directly with service providers and stores, and create a wealth of data through these interactions. As in the other cases above, the data network effect would run at the core of how these groups would then grow and continue to add new users.