Part 1: What is virality?
Part 3: Building a virality model
I fully agree with the prevailing wisdom of the marketing community regarding virality, which is that extreme virality is largely dependent on the core design of a game (or app). But virality as a characteristic of a game isn’t strictly binary; a game’s virality can be described by a point on a spectrum, and I’d argue that viral mechanics can be engineered which move that point in the direction of higher virality.
The classic example of an app that exhibits fundamental virality is Skype: as a communications platform, a user can’t utilize the service without inviting his or her friends into it. But that virality only really applies to Skype’s free voice, video, and chat products — it doesn’t apply to its paid Skype-out or Skype enterprise products, which are not built around the concept of a user accumulating a network of friends. And with these paid products, Skype does implement viral mechanics — through Facebook connect. This is all to say, while an app with core functionality predicated on network dynamics is viral by nature, there’s no reason that apps that aren’t can’t achieve some virality.
Angry Birds is a perfect example of this: the original app wasn’t inherently viral, yet it achieved 50 million downloads in its first month. Had the Rovio team concluded that an app without a fundamentally viral core concept (i.e. platform that requires the invitation of friends) couldn’t achieve significant growth outside of paid acquisition campaigns, the game probably wouldn’t have been undertaken.
A viral concept such as PVP / co-op missions as the fundamental component of gameplay is obviously the easiest way to engender significant virality. But virality can be achieved with non-core components that encourage sharing and make that sharing effortless; a basic level of virality can be achieved without creating a platform.
So how can viral mechanics be engineered? Like any other gameplay feature, through measurement and iteration. K-factor is a function of invitations and conversion (accepted invitations), so optimizing the volume — and, more importantly, the quality — of invitations sent by users will drive a game’s k-factor up. Note that measurement and iteration isn’t the same as plugging a Facebook connect button into a game and then pushing spam into a Facebook user’s news feed; measurement and iteration means devising a viral mechanic and honing it through observing player behavior.
Perhaps measurement and iteration won’t increase a game’s k-factor by more than a few percentage points. How do we know if those few percentage points were worth the time it took to engineer the viral mechanic — in other words, if the opportunity cost is justified? By modeling virality and understanding the value of our users.