If you’ve gotten this far, it’s probably safe to assume you understood all (or most) of the things we went over in our last chapter.
Maybe you’ve even already tinkered with the basic formula on your calculator or in a spreadsheet to see the impact virality may have on your company over the course of time.
I hate to burst your bubble so quickly, but the basic model I’ve given you so far – while correct to a degree – is nowhere near complete.
The truth is that it leaves out numerous practical items that will drastically impact user growth over time.
Before moving forward, it’s important we cover these to keep building our model.
The basic formula we’ve already talked about assumes all your customers will continue to send out invitations during each and every cycle at the same rate.
This is ridiculous.
Most of the time there’s going to be some significant variation in virality over time on a per-user basis.
A lot of this has to do with a brand new stage in the viral customer lifecycle that we’re about to learn.
But to better explain, let’s first look at an example using an actual contagion.
(If you own a gas mask, this is probably a good time to put it on. If not, I’m sure you’ll be fine.)
Let’s say you’ve got an incredibly contagious virus (for example, K = 2.5), and Patient Zero goes about their normal day.
As Patient Zero traverses around their normal daily environment, they’ll have a high likelihood of infecting most of the people they come into close contact with.
Once they begin coming into contact with people initially, Patient Zero’s infection rate will spike.
Nobody in their network has ever been exposed to this virus before, so everybody is susceptible.
However, after a few days, the rate at which Patient Zero infects new people with the virus will drop dramatically.
Most people who have already come into contact with Patient Zero have either become infected, developed some sort of immunity, or had immunity to begin with.
IF a random new person who hasn’t yet been exposed comes into contact with Patient Zero, they may still have a high probability of infection, but the frequency at which that scenario plays out is relatively low.
That is, unless something out of the ordinary happens that forces Patient Zero to break his or her normal routine, OR the people within his or her usual network break their normal routine.
In summary, once somebody becomes infected with a virus, their initial probability of infecting those around them spikes but quickly drops to a crawl after most of their network has been exposed.
Those who can become infected will become infected, and those who haven’t become infected likely won’t be.
This same phenomenon also takes place in viral marketing (minus all the potential for a horrifying and painful disease ridden death).
We call it network saturation.
Once a user is exposed to a product and sees both the core and viral value, their viral infection rate spikes.
They immediately think of the people they know who will also see the value, and they send out invites in quick succession.
However, just like a user with a virus, the infected user’s infection rate is typically limited to the size of their network.
As soon as this network has been fully exposed and has had the opportunity to be infected once or twice, the infection window ends.
As a result, virality for that infected user drops to a crawl.
This is network saturation in action.
However, just because network saturation is beginning to take root does NOT mean viral growth stops.
Each newly infected user who is exposed to the product during the original user’s infection window typically has at least SOME new people in their network to varying degrees that they can expose to the product.
In addition, as new people enter the original user’s network, they, too, are likely to be exposed.
You can also begin changing your viral loop in various ways to act as a sort of “mutation” to your virus (but more on that later).
Additionally, if you get really lucky (or very strategic) and can infect a celebrity, virality can spike in a massive way.
Not only will your product be exposed to a far larger network than average, but the social proof that comes from what may as well be a product endorsement will be such that the conversion rate on those invites will go through the roof.
In other words, it will now become “cool” to be infected.
So if you haven’t become friends with Robert Downey Jr. on Twitter yet, now’s the time.
Even if you somehow manage to make your product’s viral infection become cool – network saturation will still occur.
(No amount of Robert Downey Jr. can stop the laws of viral nature.)
Eventually, users WILL expose everyone they have the ability to touch.
As such, either those people will have already been infected or are either temporarily or permanently “immune” to infection.
This immunity may come from things like:
So based on things like network saturation and immunity, how can we factor in the overall decay of virality per user over time?
It’s pretty much impossible to accurately factor viral decay into a projection equation practice – but you can estimate it.
This all starts with the assumption that you have managed to put a very sound data collection, calculation, and interpretation strategy in place.
However, until we can realistically build that, let’s continue pressing forward.
We will build our formulas by assuming our infected users are simply running out of other people to invite due to all the normal network saturation factors we’ve already discussed.
Given that assumption, let’s use a simple geometric viral decay rate.
In other words, the per-user viral factor is reduced by exactly 50% each month.
This means that after a user’s viral factor spikes, it will quickly drop by half each and every month thereafter.
While it will never completely stop so long as they’re still a user, it will quickly drop to a very slow crawl.
So how does all this apply to our original formula?
If we sum up all these monthly per-user viral factors over time, we get the lifetime viral factor, which we will name K’ (yes, that’s K with an apostrophe).
The easy part?
As long as we make sure i and conv% are lifetime factors as well, our viral equations still apply similarly, but we can now sub K out for K’ to get lifetime statistics. Thus...
u(t) = u(0) * (K’^(t/ct + 1) – 1) / (K’ – 1)
PRESTO!
Immunity and decay don’t have to be the end of the road.
...At least not immediately.
We can reduce immunity and prolong virality by changing a user’s network to be more susceptible.
How?
By mutating the infection (i.e. your awesome product).
Grab your ooze canisters. We are about to transform into the Teenage Mutant Viral Heroes.
We’re not done decaying quite yet.
In the next chapter, I’ll present a simple model on how to better predict network saturation for your viral engine, as well as provide a few helpful strategies on how to avoid it.
SIDE NOTE: if you want to hear me talk about all things growth, startups, and inspiration, hit me up on Twitter, Instagram, and LinkedIn!