The Social Graph, Facebook, and Virality

by Jesse Farmer on Tuesday, April 15, 2008

The social graph is the web of connections between friends, family, and acquaintances that everyone has. I know my friend who knows someone who works at the company I want to interview at, so he connects us and I get a shiny new job after acing my interview. It helps me meet new people, find new music I like, and generally navigate my social world. If I find something because of the connection in my social graph I'm much more apt to trust its worth. After all, people I know recommended it!

The thing all social networking websites, like MySpace, LinkedIn, Friendster, and Facebook, have in common is that they try to create a virtual copy of the social graph. The graph serves as a way to spread whatever content the site is pushing. If the representation of the graph is good enough I'll know what my friends are reading, watching, doing this weekend, etc.

I don't think it's controversial to say that among all the major social networking sites Facebook has the best representation of the social graph. MySpace's representation is too connected — my real-life connections might be there, but I also have a million other connections that don't reflect anything in the real social graph. Friendster's representation isn't connected enough since none of my friends use it. Facebook, however, has hit the sweet spot: collecting friends a la MySpace is actively discouraged by the features on the site, but they began with a narrow enough demographic that they were able to reach a critical mass among college students.

The Social Graph and Virality

The social graph is closely related to so-called "viral" services. News can spread faster through the lines in the social graph than it could otherwise. By having an accurate virtual representation of the social graph Facebook is able to amplify the usual word-of-mouth effect. Everything I want I can broadcast instantly to all my connections. Facebook does this using the News Feed and Mini-Feed features, showing my friends' activities chronologically and my own activities in my personal profile.

Now we have the Facebook platform. Some people mistakenly treat it as a fancy "widget platform," but they don't realize that the platform has moved Facebook far beyond the tired MySpace + widgets = success formula. In addition to providing compelling content every social networking site wants to build a quality copy of the social graph, but in one fell swoop Facebook has given away access to the highest quality copy on the web, namely theirs. Every developer or entrepreneur looking to build a social networking site will have to ask themselves from now on, "Would it just be better to write my application on the Facebook platform?"

I don't think one can overstate the power of the position this has put the Facebook in, from both a business and technology perspective. If they can actually cultivate a market around Facebook apps — a proposition which is becoming increasingly likely with the creation of programs like appfactory — they might very well move into truly revolutionary quarters.

Powerful, Yes, But no Magic Bullet

But what does this mean for the app developer who just wants to create something popular? The integration with the Facebook news feed allows apps to dramatically increase their virality by piggy-backing on Facebook's copy of the social graph. But the Facebook platform isn't a magic bullet. Turning your website into a Facebook app doesn't mean it will become an instant viral sensation — you also have to understand how to use the social graph.

Let's say you have a popular blog and want to increase people's interest by somehow integrating with Facebook. Your first inclination is probably to let users add a feed of stories from your blog to their profile. Even if your blog is popular this idea probably won't do much good for two reasons: one, readers will have to leave Facebook to get the content, interrupting whatever they were doing; two, there is no social component.

The first problem is probably not worth your time to solve if all you're doing is recycling content from your blog. The second problem is more fatal because it runs against the idea of the social graph. The content being served is in no way related to the person whose profile I'm viewing. It's not stories they've commented on, stories they've recommend, or anything of that sort, it's just a boring list of all stories. Unless there is a social component do not expect Facebook to miraculously drive 30 million users to your website.

Actually, don't expect it to drive traffic to you site at all. Like I said, Facebook users don't want to leave Facebook. As a concrete example, I wrote the PopSugar 100 application for Sugar Publishing. It drives about 5K visits per week, which is respectable, but a drop in the bucket relative to our usual traffic. You can try to monetize the application directly on Facebook but if you're a medium-sized startup already monetizing your userbase in other ways then a Facebook app probably best serves as a way to increase your valuation.

Case by Case

Edit: Joyce Park of Renkoo has written up a guide about designing for the Facebook platform that hits a lot of the points here. You should read that, too.

To see that the Facebook isn't a panacea for your viral distribution problems let's look at some case studies.


    Users per hour

    The digg app is an example of an application which you would expect to do well on Facebook since it has a huge social component but which is essentially floundering. Where other apps with much less impact than digg are growing at a rate of hundreds of users per hour, digg's app is almost static.

    This is because, ironically, they don't harness the social graph at all. All the app lets me do is post a list of stories I've dugg into my profile. I've suggested five ways the folks at digg could improve it. The easiest way is to post all news items I submit or digg, at my preference, to my news feed. Not only does this benefit digg by bringing in more users but it makes digg much more useful because I can promote my stories on digg through my corner of the social graph.

  2. Zombies

    Users per hour

    This app is an example of an application which is entirely viral but provides little content. As you'd expect, it spread very fast initially but is becoming saturated. I suspect, as time goes on, people will uninstall it because there is little there to hold people's interest. Being viral helps your application spread but if there is no substance to back it up people will eventually leave. We'll see if the authors of Zombies can create compelling content to keep people involved.

  3. Graffiti

    Users per hour

    Graffiti is probably the perfect example of a Facebook application right now. Not only is it extremely viral, it augments an existing Facebook service so there is essentially zero learning curve. I'm informed when someone's Graffiti wall is updated via the news feed and people can share particularly clever drawings with each other.

    It also shows the power social networking software can have in our real lives. So many friends of mine have the Graffitti Wall that it has started seeping into my real social interactions. Universal access among people connected to me means that it becomes a useful avenue of social expression. More plainly, when something amusing happens I might think of a great thing to draw on someone's Graffitti Wall.

  4. Booze Mail

    Users per hour

    Booze Mail, like Graffiti, does a great job at exploiting the social graph. I'm informed whenever a friend of mine sends someone a drink, so not only do I learn about Booze Mail, but I get a better glimpse at my friend's social network. I'd bet money that Booze Mail is going to be one of the million+ user applications within a week. Indeed, of the 20 apps posting larger absolute per-day gains, only three have less than a million users.

  5. Matches

    Users per hour

    Matches is one of a dozen-or-so apps posting regular hourly losses. It serves as an example of the pitfalls inherent in the Facebook platform. If you look at the Matches discussion board you'll see that, for whatever reason, it isn't working for most people. Part of the problem is that Facebook's model requires the app developer to shoulder all the load.

    If you created a popular app then you'd best be prepared to deal with Facebook-sized traffic. If you can't then people will get fed up. Apps like this show that people are definitely willing to uninstall applications that don't do what they want (regardless of whose fault it is). So even once you have a userbase you're not guaranteed to keep it.


The Facebook platform has a potential to revolutionize the way we think about the social web. It serves as a gateway to a high-quality copy of the social graph that exists in real life. This graph, in turn, lets people share content, ideas, money, goods, and all sorts of things with ever-increasing efficiency. The degree to which the graph permits targetted selection, e.g., find me all influential people who like to read between the ages of 20 and 25, is an advertiser's wet dream.

But the platform is no magic bullet. Applications you think would do well sometimes falter and you're not guaranteed to be a viral hit just because you've created a widget for your blog. Apps are rising and falling all the time and the market is still taking shape. It's an exciting time, regardless.

All stats brought to you by Appaholic.