Algorithmic Lock In Effects

I view companies and entrepreneurs through my own lens, and have been developing my investment thesis for some time. Recently I have been thinking more about the software (vs. the wetware). Specifically, thinking through services that use algorithms as a “lock in” mechanism. This started at the end of last year when I received the incredible summary from Spotify that shows minutes, songs, artists and more (see the official Spotify Wrapped if you use the service). This is a great example of algorithmically locking me into the service.  The more I use it, the more interesting data (or value) the service can return back.

So, can a algorithm keep you locked into a services? To answer this question I was thinking about all the services that actually become better the more you use them. Spotify is a good example, as is which “scrobbles” (my listening history for 10 years at the right) your music from every other services, arguably making their recommendations better. I believe the same is true for Netflix, which uses ratings and reviews and obviously watching habits to keep making better recommendations. In fact, every action within the service probably is an indicator of what you want to watch, when you want to watch it, and how to deliver more value. The development patterns at Netflix reflect this with the advent of “skip intro” and auto-playing a movie/show when you hover over a choice – keeping you constantly engaged with the service. The other big example here is Facebook, with its now famous algorithmic newsfeed change. The same is true for Instagram. Gone are the chronological feeds of yesteryear – probably proved with data from product teams that show activity and engagement jumps because of the change. I am sure there are countless other services that exist that do the same. Some of the less loved include banks that have high switching costs to get off their platform. This arguably provides no value, but does have the intended effect. 

What is most interesting about platforms that have a data lock in (no export capability) thinking they are keeping customers from churning are actually doing themselves a disservice. It turns out that platforms that have an enterprise client requirement to have data portability and archive or export capabilities have a happier customer base because they know they have the “option” to move but may never do so. A good example of this is the Google Takeout service that lets you export and take your Google data, including all archives anywhere. The interesting fact is that it’s hard to find a better free provider than gmail, but you are trading some of your privacy and data exhaust towards Google. I have seen a handful of enterprise examples where large clients didn’t want to be “locked in” to a platform without future portability – upon building this out they signed up, only never to use the feature.

This brings me to exploring the thesis that systems that return value based on continued usage compound the value to the user over time. Although dark patterns exist that abuse this, when used for good they definitely deepen the competitive moat of a platform or service.

Algorithmic Lock In is another example of data exhaust being used effectively.

In a world that is moving towards distributed and decentralized platforms, those experiences that get better over time definitely keep you on that service or platform.