So you want to optimize the virality of your app. What metrics do you pick? There are dozens of numbers to look at. If you’re feeling confused and overwhelmed, don’t panic. I’ve been there, having helped run growth at Dropbox where social invites through referral and shared folder invitations were a huge part of our growth. (psst, now I’m working on YesGraph)
You might be surprised that I’m not going to talk about “Viral Coefficient” or “K-Factor”. They lack a strict definition (hint: time bounds matter) and aren’t actionable. I’ve also noticed that people that talk in these terms often haven’t worked for a massively successful company. So don’t be that guy.
At the start, keep it simple. Here are the three metrics I like using.
- % users participating
- # Invites sent / user on average
- # signups / invite
Let me try to define each.
1. % Users Participating. At the start, you want to know how many people are even participating in the invite flow. This doesn’t mean they’ve sent invites, just that they are actively involved in the process. For example, visiting a referral invite page like dropbox.com/referrals could count.
I like this metric especially because it is usually surprisingly low. Many sharing features only have 1% or 2% of users participating in a given month. You can cheat a little and only measure the percentage of active users. That doesn’t help much, but it does make your numbers look better. Try to make the events meaningful. Just being on a page with a small “share” button shouldn’t count.
2. Invites Sent / User Participating. This can get tricky depending on your exact flow. You might, for example, want to track percentage of people that publish to Facebook (which is one to many) separate from email invites (which are one to one). Email invites tend to perform far better (10X conversion) than publishing a link to Facebook or Twitter, so it is ok to focus on how many are sent. If publishing dominates over invites, add metrics for % participants who publish and # clicks per published link.
3. Signups / Invite. This measures how many people signup per invite sent. Again the math can get tricky if you’re looking at published links vs direct invites. Conversion of an individual signup will be less than one. From publishing a link you can get multiple signups.
You can further refine each of these measurements. I highly recommend you get started with the basics to understand what is performing well and what isn’t.
At the top level you might attribute X% of your signups to a sharing or invite flow. But if you want to make that number go up, you need to know where the dropoff is.
With Yammer, the last time I checked, everyone that signs up is presented with an invite flow to get 5 people into the service. I don’t typically recommend prompting invitations before engaging the user, but some social apps require others before becoming compelling. Some of the invites might be low quality for people that just want to get in. This means the conversion to signup will be lower than expected.
Here is a guess at Yammer’s stats:
- 100% of the users see the invite flow.
- 1.5 invites per user.
- 0.10 signups per invite.
The only reasonable place to focus your effort is on conversion of invitees. 10% has a lot of room to grow. Here is where this gets actionable. Run tests on the messaging. Send multiple invites. Run tests on the landing page. Prompt the inviter to send reminders.
Now take a mythical service that just started making a referral program, InstaJoy. The only user exposure to the referral program is a link at the top of the home page that says “Give $10 Get $10”. Sounds compelling, but what percentage of users are looking at the home page at any one time? And how many even click that. This is an example where not enough people are participating.
- 2% of users see the referral page.
- 0.5 invites per user
- 0.25 signups per invite
In this case, what might happen if the service sent an email to all users to make referrals? Maybe 2% of people would click the call to action in the email. But that means ~4% of users see the referral page, effectively doubling performance. If they send another reminder email, another 1% might engage. That is another 25% increase in performance.
These metrics are simple yet actionable.
Again, you can get a lot deeper in every dimension. Don’t let perfect accuracy get in the way driving immediate improvements.