Virality rarely just happens — like any good growth channel it needs to be engineered. The first step is mapping out your viral growth loops. What are the different stages a user goes through from experiencing your product for the first time, to participating in the sharing of your product. The output of this process should lead to the input for the beginning of the loop.
To show how this works, let's map out a classic example of a viral growth loop — social networks.
When we have a model like this built for the viral loop we plan to use to grow our business, we can make better decisions about what to optimize and model out its potential impact, as well as forecast future growth numbers. Let's go through each part of the model one by one, and define each metric. If you're interested in diving deeper into actually building a model of these calculations, check out our post on Modeling Viral Growth.
When you sign up a new user, how many of those users invite their friends or colleagues? The core focus of the Facebook growth team was getting new users to 7 friends in 10 days, as this increased both virality and retention — Facebook is useless if your friends don't use it. For a social network this metric is key, as you can count any user that doesn't invite friends as lost.
In our model we assumed 75% of users participate in sharing with friends. So if we have 100 new users, then 100 x 0.75 = 75 of them will invite friends.
When someone invites their friends or colleagues, how many do they send? Some might invite only 1 or 2 close friends, others might spam their entire address book. For Facebook it was 7 friends, but the average will vary depending on the product, audience and how well optimized the invitation flow is.
For this model we assumed every user that participates in sharing, invites 7 people on average. In this case, with 75 people inviting friends, then 75 x 7 = 525 people will receive invites.
If someone receives an invite to join, how many people click through to the website or app? This percentage will be highly dependent on how compelling the value proposition is, as well as their relationship to the friend or coworker that invited them. For example it would be high for a social network where you're invited to talk to a close friend, but much when someone you went to high school with invites you to play Candy Crush so they can earn in-game items.
In the model we assumed a 50% clickthrough rate. This means that for 525 invites, we'll generate 525 x 0.5 = 263 clicks, or new visitors.
When a new visitor arrives on your site, what's the probability they sign up to be a new user? Conversion rate optimization is a whole discipline in itself, and practitioners know that by removing friction and using psychological tricks, you can always find a way to get more users to convert. This is also dependent on trust, both of the platform they're signing up to and the friend that invited them — to some extent one can substitute for another.
For the model we assumed a 40% conversion rate for our social network. From the 263 new visitors we drove through invites, we would expect to sign up 263 x 0.4 = 105 new users.
Finally, the last metric that factors in all proceeding metrics. Also called K-Factor (or in epidemiology, R0 or R-naught), this is a measure of how many new users you get through sharing, for every new user you acquire through other means. This is the primary measure of virality, and its importance can't be understated. If you manage to push this number above 1, you get more users in each cycle than you started with, meaning you will start to grow exponentially.
You can calculate this metric by multiplying all previous metrics. So in our model 0.75 x 7 x 0.5 x 0.4 = 1.05. Alternatively you can divide the number of new users you get at the end by the users you started with, so 105 / 100 = 1.05.
In practice a Viral Coefficient of more than 1 is extremely difficult to achieve and sustain for any length of time — if you're lucky enough to 'go viral' you often saturate your audience and see a rapid decline when there's no one left to sign up. However every basis point of virality is valuable to your business, because it makes every other marketing channel perform more efficiently.
For example imagine you were paying $10,000 to acquire 500 people on Facebook, a $20 cost per acquisition. Increase your Viral Coefficient to 0.20 by encouraging sharing, and those 500 people will bring 100 more, who themselves will bring 20, who bring 4.
This downstream impact of new signups is called the Growth Multiplier, and you can work it out by dividing one by one minus the Viral Coefficient. So a viral coefficient of 0.2 would give you a growth multiplier of 1/(1-0.2) = 1.25. This means that every new user you drive from Facebook ads is worth 1.25 users — you can afford to pay 25% more for user acquisition.
Note: this formula doesn't work for Viral Coefficients higher than 1, because they trend to infinity. Instead you must choose a time period to predict, or number of cycles into the future. The formula for this initial users x (viral coefficient ^ (cycle + 1) - 1) / (viral coefficient - 1). So for our viral coefficient of 1.05, to see 10 cycles into the future this would work out to be 100 x (1.05 ^ (10 + 1) - 1) / (viral coefficient - 1) = 1421 total users.