Why Online Games Are Tanking

I recently finished work on an algorithm that draws a predictive curve showing user engagement over the lifetime of a game. It took me eight years, and the more I play with it, the more it teaches me. I'm not here to sell my algorithm, I plan to make it open source later this year. What I am here to tell you is that we have been approaching online games in a very ineffective manner.

Let me start by using a movie analogy. When I want to see a movie, I can look up its critic ratings and get an idea how the movie was... at the end. What I mean is the critic presumably watched the whole movie then wrote the review. In a game this is a bit tricky. When did the reviewer write the review? At the end? How long did this take her? Did she do all of the optional content? What if the game has no “end”?

In the context of online games, which rarely have an “end”, ratings are meaningless unless you state exactly when in the game you gave the review. Your impression five minutes in might be totally different than it would be five hours in. If you are trying to monetize your game the same way at minute five as you are at hour five, chances are you are losing a lot of money.

Now what if the designer has no concept of how time affects their product, but just knows that user engagement drops RAPIDLY after the third hour. Thus they try to monetize intensely in the first three hours. What if the reason the engagement dropped so rapidly in the first three hours was the method of monetization itself? This becomes very difficult to detect and can lead to a “chicken or the egg” kind of paradox.

It has become a common myth that Free to Play games have rapidly degrading user engagement. This has led to ever more aggressive means of monetization, which actually lead to even faster rates of engagement degradation.

The problem here is that we are still thinking that our online games are retail products. If I was trying to sell you phone service, would I try to aggressively monetize you in the first three hours before you canceled your service? We are not building games as if they are services, and I think this is largely because we do not know how.

Why is it that EVE Online has user numbers that just seem to go up every year? Why is it that midcore games have user numbers that drop by at least half on every server every week? This is not because of the graphics. It is partly due to the way these games are monetized, but that is a relatively small component of the effect.

I think we are afraid of our consumers. We know they will betray us and we try to “get what's ours” before they leave us. This is a destructive relationship. The key to maximizing user engagement, and of course monetization, is to provide a product the user will embrace. You will have to wait for my book to see how I solve this problem, but I want the community to start thinking along these lines in the meantime.