What is a good way to measure game virality?

Question

I have added some social features to an iPhone game (Lexitect if you're curious), such as email, Twitter, and Facebook integration for sharing high scores. Along with these features, I am measuring how many times users make it to each step. The goal of these features are to make the game more viral, and I am trying to get to a measure of game virality.

I would think that a game virality metric would produce a number based on 1.0, where 1.0 = zero viral growth, and 1.01 would represent 1% viral growth over some unit of time.

How is virality normally measured, and in what units? How is time capped on the metric? i.e. if I gave each player a year to determine how many recommendations they make, I wouldn't get any real numbers for a year from the time I start tracking it. Are there any standards for tracking virality in a meaningful way?

Answer 1

Virality is usually measured by it's K-Factor. The two links in the wikipedia article there are helpful as well - the article on viral app objectives lays out things, and Jon Radoff's spreadsheet can be interesting to play with to figure out some of the interconnections between various factors.

K-Factor represents the lifetime number of infections that a single infected person makes. So, if your K-Factor is 1.1, then each infected person will infect 1.1 other people - in other words, you are headed for exponential growth. K-Factor does not include the mean time to infection, that's a separate (but obviously important) item.

Truly measuring virality, from what I understand, can be difficult because it depends on your metrics tracking being in place in the right spots, and then you have to keep and correlate a lot of disparate user data.

Answer 2

So, I'm not sure what you mean by "virality". I'm going to assume you mean, "I want to track how often my players recommend my game to their friends and track how many of those recommendations turn into conversions."

Check out Google Analytics and Omniture. They provide a great way to track stats across a variety of platforms and also provide ways to visualize those stats. Really, though, you could roll your own without too much effort.

Let's take one case: Player A downloads your game and starts playing. As part of being run for the first time, your game generates a unique ID.

Now suppose Player A is playing your game and it sends out a "I reached level 12! Play with me!" notification to the social network du jour. But wait! Each link embeds another unique code in the link to your homepage from which players can buy your app -- via an iTunes link or whatever. This second unique code includes the first (which uniquely identifies a player) plus some extra stuff (what level they're at, etc).

Along comes some of Player A's friends imaginatively named B, C, and D. Each one of them clicks this link and is directed to your homepage.

You can then see if this person clicks the link to buy your app. By tracking the "arrived at your homepage with unique code" numbers versus the "bought your app" numbers, you could come up with a rough, continuous measure of how often you're converting random people to game players.

EDIT: By the way, if I'm completely misunderstanding your question please comment and I'll pull this answer.

Answer 3

The only way to really test it is to measure the outcome of those messages. Eg is there a way to correlate the number of tweets to the number of new downloads.

Speaking purely about twitter, the easiest way to measure virality is to see how often a tweet gets retweeted.

For example, I have a website whendidyoujointwitter.com - it has a 'tweet this' type of link. Often when someone tweets that link, some of their followers will go the link in their tweet and in turn tweet their own message. We can then measure the virality if we see on average how many followers do this 'retweet'.

Eg if someone with 1000 followers tweets it and then 10 of their followers tweet it as well, this is a success rate of 1%.

Now, this is still not a virality percentage. We have to go a step further by asking how many followers does the average twitter user have? If everyone had 1000 followers, then those 10 retweets would result in 100 retweets which would result in 1000, etc (until it peters out - there are only so many twitter users/people in the world) - ie very viral by most standards.

BUT not all twitter users have 1000 followers. What we can try to calculate is a tipping point. With a success rate of 1%, how many followers would the average user need to have in order for the meme to keep spreading?

If the average is 100, then 1% means a tweet would result in 1 more tweet - ie the meme would be kept alive. If the average is less than that, then the meme is sure to die out.

Now, going in reverse, the latest stats I can find show the average follower count is 70. So a success rate of 1% is not enough to go viral - it would need to be up around 2% or above.

Uhhhhm so thinking about, I guess it's a logarithmic/exponential thing.

• 100% success rate (all retweet) = viral maximum
• 10-99% = extremely viral through this whole range
• 2% success rate = viral minimum
• 1% or less = not viral