Add me to your network: How LinkedIn hacked growth by creating viral loops

The following is an edited excerpt from Startup Growth Engines: Case Studies of How Today’s Most Successful Startups Unlock Extraordinary Growth by Sean Ellis and Morgan Brown. The excerpt was provided by the authors. You can buy a copy of their book here.

The value of active users for LinkedIn isn’t just in their engagement – active users also play a critical role in both re-activating inactive users and bringing in new ones, making virality one of the biggest keys to LinkedIn’s growth.

It’s easy to understand why. A network of professional contacts is only as valuable as the richness of the results when you’re searching for a connection. In order to make LinkedIn the de facto professional network Hoffman knew they needed to get as many people on the site as possible.

Hoffman claims, “2003 was all about tuning and viral growth.” He continues: “We had this initial challenge of, ‘How do you get a million people?’ The first challenge was getting enough people so that functions like searching for people or sharing information had enough people in it to be valuable.”

Early employee Josh Elman credits the Double Viral Loop for LinkedIn’s early growth. This virality wasn’t arrived at by accident, but actively tweaked, tested, and encouraged. As former LinkedIn employee Keith Rabois echoed this sentiment on Quora, explaining, “LinkedIn also intentionally deferred any features related to revenue or engagement until after the growth path was established, which took nearly 1.5 years.”

So what exactly did they do to encourage this virality? The more apt question might be, what didn’t they do?

Photo credit: Ben Scholzen.

For starters, they tested everything. From the number of friends new users were asked to invite to the invitation messaging, the company worked to find the winning combination that converted the most invites to new users.

For example, they found that four was the magic number of default email fields for the invites page – any less, and people simply skipped over them; any more, and people got overwhelmed.

Similarly, the seemingly “throwaway” line at the end of the LinkedIn invitation copy below was shown to result in more invites sent by new users: “So I found you while I was looking around the network. Let’s connect directly, I’m happy to help you with requests and forward things incoming. It will probably make both of our networks bigger.”

Elman refers to this as inception, or planting the idea of why your product is going to be useful or meaningful to users before they ever even sign up. As Elman explains, these combined efforts led to a decent but slow level of virality, or a single viral k factor.

The vast majority of users had a single connection – meaning they had accepted an invitation but not gone through the steps to connect with or invite anyone else. Less than 25 percent of new users actually went through the process of typing in email addresses to invite new users. It’s at this point that LinkedIn began dabbling in the concept of contact importing, something that’s commonplace now but was pretty novel in 2004.

At this point in time, the concept of typing in your email and password and letting a website sift through your contacts was new and furthermore most people didn’t even have contacts in their webmail clients.

To account for these challenges, LinkedIn built an Outlook plugin that could be installed and mined a user’s Outlook contacts for them. Though the process was certainly laborious by today’s standards, it was still a vast improvement over manual entry.