Unlocking Viral Secrets On Facebook: MIT Study

A rare randomized study on Facebook by MIT shows how different marketing strategies perform.

Unlocking Viral Secrets On Facebook: MIT Study


Brands can increase the number of people who install their Facebook apps by 400% through a combination of private and newsfeed messaging, according to a new MIT study. The effectiveness of different Facebook strategies was determined by randomly giving users of Facebook apps different experiences and observing how activity spreads throughout the network. Researchers found the winning strategy to be up to two times more effective than email, and 10 times more effective than banner ads. Here’s how it all breaks down.

Passive Vs. Active

After gaining rare access to an actual business about to launch a Facebook app, Professor Sinan Aral of NYU and MIT and Dylan Walker of NYU were permitted to randomly assign two versions of the product and record user network activity. (The brand, whose name has been anonymized, created an entertainment app where users share information about movies).

In the “passive” broadcast version, the app would automatically post to a user’s newsfeed as they interacted with the product, such as declaring their favorite movie or writing a scathing movie review. In this version, new user adoption jumped 246%, compared to a version in which broadcast messaging was disabled.

For the “active-personalized” version, which combined passive broadcasts with a prompt to send personalized direct messages to friends (such as, “Hey! check out my new favorite app”), overall adoption was 344% higher (or a 98% bump over the broadcast version). Thus, adding a personalized message feature substantially increases adoption (three times more effective per message), but because fewer users will bother to send the message, adding personalized messages produces less of a bump than passive alone.

Aral also tells Fast Company that passive messages are the quickist way to drive adoption (3.17 days to first adoption, compared to 4.77 days for personalized messages), but users will be 7% less active with the product. “Passive features are better for spreading a product widely and for quick adoption. Active features, on the other hand, are better for building a loyal customer base and make a product stickier,” Aral says.


Email Vs. Banner Ads

Aral’s viral strategies are up to 10 times more effective than banner ads in converting users and around twice as effective as email advertising. He also writes that “notifications and invites also vastly outperform the ad campaign used in our recruitment phase on Facebook,” meaning that, for this particular product, there was little need for a traditional banner ad campaign to generate an existing base of users.

However, the paper cautions that for some products, especially those with small user bases, some paid ads may be necessary to shore up a group of early adopters.

Deeper Insights

Viral strategies have consequences beyond mere adoption. Active-personalized messages increased the frequency and duration of use. So, for instance, if Barack Obama or Mitt Romney are looking to secure the most amount of committed users inside of a campaign app, personalized messages might be the way to go. If, on the other hand, a movie trailer app just wants to spread awareness, broadcasting-type messages would seem more lucrative.

Trade-Offs and Experiments


Aral tells Fast Company that there’s a trade-off between broadcast and personalized messages. “The cost of passive features is the annoyance factor of spammy interaction,” he says.

Brands should be weary of the unmeasured impact of clogging up users’ news streams with advertisements. This might not mean much for an unknown startup just trying to get attention, but a brand like Coke or Pepsi has to think long term about how they are perceived.

Facebook has since blocked some of the broadcast powers of applications to rein in the spam, but app developers such as Zynga have found workarounds).

Given that social media policies and platforms are ever-changing, Aral’s study is perhaps most useful in thinking about how brands themselves could experiment with viral strategies. Randomizing versions pinpoints what works, what doesn’t work, and how to improve–instead of choosing one strategy and praying it was better than all of the others.

Follow Greg Ferenstein on Twitter. Also, follow Fast Company on Twitter.

[Image: Flickr user eviltomthai]

About the author

I am a writer and an educator. As a writer, I investigate how technology is shaping education, politics, Generation Y, social good, and the media industry.