Virality for Zombies

I briefly discussed virality in my previous post, but I thought this deserved a bit more detail. Most of what you need to know can be found in David Skok’s virality blog post, but it’s quite complex for someone brand new to the concept.

Virality is best explained through example. Say there’s a zombie virus outbreak and the President of the United States asked you to predict how fast the zombie virus will spread. There are four main variables you would need to measure:

1. How many zombies are there currently? (current zombies)
2. How many people does each zombie attack and infect on average? (avg. people attacked per zombie)
3. For every person a zombie attacks and infects, what is the % chance the person will survive and turn into a zombie? (infection rate)
4. How long does it take for a person to turn into a zombie? (zombie transformation time)

The first thing to calculate is your k-factor, which is just #2 * #3. If your k-factor is under 1, then there isn’t a huge worry – the zombies aren’t likely to organically spread everywhere. If it’s above 1, well… panic time. You’ll see why in a moment by looking at the formula.

To predict the number of zombies at some time, you just need a magic formula that pulls everything together and allows you to calculate the number of zombies at time t. Let z(t) represent the number of zombies at time t, where t is measured in days.

z(t) = z(0) * (K ^( t/r + 1) – 1) / K – 1.

z(0) = current zombies

K = (avg people attacked per zombie) * (infection rate)

r = zombie transformation time

Simply plug in a number of days into z(t) to predict how many zombies you’d have on a given day.

If you wanted to reduce the rate that the zombies spread, you would need to figure out ways to reduce your k-factor and increase the zombie transformation time. Perhaps you could develop a drug that would relieve zombie symptoms and delay zombie transformation time. Or, perhaps you could quarantine the zombies to limit the avg number of people each zombie can reach.

Congrats! You now have recommendations for the president that could save the world.

More importantly, you’ve also just learned how virality works for internet applications. Except, instead of trying to reduce the rate that zombies spread, you are the villain that tries to increase the rate that the zombies (your customers) spread.

The mathematical model is exactly the same, with some slight name changes:

c(t) = c(0) * (K ^( t/r + 1) – 1) / K – 1.

c = number of customers

K = avg number of invites sent per customer * avg. acceptance rate per invite

r = time it takes for an invitee to turn into a customer and send her invites

The variables you have control over when trying to increase the virality of your app are K and r. The first goal is to get K as high as possible, ideally over 1. The K formula defined above isn’t very granular, so it may be more useful for you to break the K formula down into multiple steps based on your sign up flow and invite loop. One complication is having multiple invite flows (e.g. users can invite via FB, email, and Twitter). In this case, I’ve found that you can simply calculate the k-factor of each invite flow and add them all together to get your total K.

The second goal is to lower r. You do this by making your sign-up and invite flow as easy and seamless as possible. For instance, you could reduce the number of steps, take out unnecessary fields, or experiment with stronger calls to action.

It’s difficult to have an app that is truly viral, so it may be that your app is not inherently viral and will never reach 1. In that case, your best option is to maximize your K and combine your partial virality with other marketing channels (SEO, SEM, etc.). By increasing your K, you are also increasing the value you get out of your other marketing channels because (as Skok mentions) you increase the overall number of customers gained per lead you acquire from each channel.

Enjoy exploring the world of virality. You’ll know who to save first when a zombie apocalypse arrives .