Fall 2015 marked the beginning of my long-term project involving analysis of PC games released on Steam. My dream goal was to provide a tool which would tell developers how unique their game is, how well it is going to sell, what would be the ideal pricing. Does it sound completely unrealistic? It should. Video games are highly unpredictable, and how to make sure a game gains popularity remains quite a mystery.
Why would anyone even bother with doing this, it’s predestined to fail from the beginning. And that’s exactly why I started, I don’t get scared off so easily. I’m not entirely sure what the outcome will be, but something useful is surely hiding within the pile of data Steam holds.
So I started with goal #1: Investigate how predictable the success of games is based on the info available in the Steam catalogue before release.
But that’s crazy! You can’t predict anything based on that! Or so I heard. Still, it seemed like a pretty good start. What determines a successful launch? A lot of good reviews? Social media hype? Sounds likely. Basic metadata like genres, features? How could this even… , that’s not how it works… OK, let’s do it!
I collected catalogue info from August 2012 to July 2015 (including). I decided not to include free-to-play titles and titles in Early Access. The dataset consists of roughly 3,000 games from the Greenlight era. It contains attributes such as required age, languages, genres, game features (multiplayer, co-op etc.), Steam features (trading cards, Workshop, achievements), number of screenshots and trailers, description, release date, launch price, and more. No reviews, no scores, no social media numbers.
I also added a simple measure of post-launch success but I want to talk about that later. Let’s start with a nice simple graph showing the number of games released each month. As you can see below, Greenlight gradually allowed a lot more games to be released on Steam than it was usual before.
Does it mean there’s now a lot of successful games? You might have already heard that Steam has mishandled curation of their library. In the next post, I’m going to add some color to the graph in order to find out how games are doing after launch.
Spoilers: it’s pretty bad.