Skip to main

3 Scientific Studies With Real Insight Into Social Media

Authors: Pratik Dholakiya Pratik Dholakiya
Posted Under: Social Media
Listen to this Article


badge-guest-post-FLATTERIt’s no secret: social media marketing has a data problem. Despite a growing number of tools available to measure the ROI of social media, many of the claims made by “marketing gurus” in this sector come from the gut. Few things are more refreshing in this industry than hard evidence, and that’s why I love looking at what peer-reviewed, scientific studies have to say on the manner.

It turns out social media may be even more influential than we have imagined, just not in the way many of us have envisioned. Let’s take a look.

1. Sales and Facebook, Only the Interactions Matter

Social media marketers have been arguing for a while now that social networks aren’t useful as a “broadcast” medium. Very few brands have taken this advice to heart, opting instead to find ways to extort Likes out of their audience, and use that as an opportunity to bombard them with spam.

The “edgier” social marketers who tout “relationships” and “connections” will probably find themselves feeling vindicated after taking a look at a recent joint study by the National University of Singapore and Nanjing University.

A team of three professors worked together with a small Asian retailer to investigate how Facebook influenced sales. And this was a case study if ever there was one, because the data they had to work with was phenomenal:

  • A database of 14,000 customers, including, get this, actual sales data
  • Facebook API data
  • Collaboration with the Facebook Data Science Team

And instead of measuring some meaningless metric like the number of Page Likes or something, they measured which customers saw which posts and comments, they used text analysis to quantify the information richness and emotional sentiment of them, and they categorized them into marketer-generated versus user-generated, and direct interaction versus indirect interaction.

And these guys aren’t amateurs. They performed all kinds of statistical voodoo tricks to control for the possibility that people were joining the Facebook page because they were already going to spend money, not the other way around.

The end result was definitive. On average, if you joined the Facebook Page, you spent about $22 extra on the company. More importantly, the difference was entirely explained not by the simple act of joining the page, but by the interactions that took place on the page.

Here is what they learned:

  • Broadcasting on Facebook really was pointless. We can’t rule out the possibility that this particular fashion retailer just wasn’t skilled enough to use “broadcast” messages on Facebook to generate sales, of course, but that was the result. When it came to interactions with the marketer, only direct interactions influenced sales.
  • It’s probably not surprising that when it came to these direct marketer-user interactions, only positive interactions boosted sales. It might be surprising that the information richness of these interactions didn’t seem to matter.
  • All in all, user-generated content influenced sales far more than anything the marketer did.
  • Paradoxically, the most influential user-generated content was positive and indirect, at least when it came to sales.
  • Direct user interactions were very important, however, because they transformed the products into inelastic commodities. (The elasticity of demand could get as low as 0.006, for those who speak the lingo.)
  • Whether user-interactions were direct or indirect, information richness had a positive impact on sales.
  • When it came to direct interactions between users, only information richness influenced sales. The positive or negative tone of the interaction didn’t seem to make a difference.

Needless to say, these results demonstrate that the real world can be a convoluted and unexpected place, but we can shorten this down to a relatively small list of takeaways:

  • If you can do only one thing on Facebook, encourage information rich interactions between your users, both direct and indirect.
  • If you can do two things, interact with users directly, and stay positive when you do.
  • Recognize that while positive, indirect interactions have the strongest influence on sales, all information rich interactions help improve sales, and both direct and indirect interactions play crucial roles for different reasons.

A good example of a business that’s pulling this off is Shopify, as this post demonstrates. Rather than posting links to their new POS hardware, you see them asking questions to get people talking, and responding directly to the comments. This is the kind of Facebook behavior that leads to sales.

Now, it’s important to recognize with studies like these that whenever we say something didn’t influence sales or didn’t have an impact, we can’t necessarily generalize that. It’s still possible that broadcast-style messages could influence sales in some circumstances. It’s still possible that positive comments could hurt sales, or that it’s better to be information rich when you’re communicating directly with your audience.

One thing that stands out to me, here, is that the brand is using Facebook alone to influence sales. It isn’t trying to drive referrals to a site or get anybody subscribing to a blog. There’s no all-out content marketing going on, so I wouldn’t rule out the fact that you can influence sales with indirect communications. But this study does make it clear just how powerful user interactions are, and why they might even be more important than anything you produce on your own.

So, how do you build this audience of users to interact with each other? Is it with “shareable content?”

2. Viral Marketing? Yeah, About That…

It seems like every marketer and their grandparents are telling you that if you want to succeed in the world of social media, you need to produce content that is “worth sharing.” I’m certainly not going to all-out disagree with that premise, but I am going to say that if you think it’s top priority, this next study is going to shatter a few preconceptions.

The study is called “The structural virality of online diffusion,” a team effort by Microsoft Research and Stanford University, led by Sharad Goel. In this study, they analyzed events as they spread through Twitter, in an effort to understand how and why topics start trending on the platform.

They looked at roughly one billion events as they cascaded through Twitter.

And we all know how this works, right? The most popular tweets get shared with friends of friends of friends. It’s that six degrees of separation thing, right?


True, they did find that the most popular events on Twitter were in fact more viral that they average Tweet. But the correlation between virality and popularity was surprisingly low: 36 percent. It’s impossible to predict how viral a piece of content is based on how popular it is, and even the most viral events could see audiences of just a few hundred.

More importantly, the team of researchers didn’t find a single example of a “supercritical diffusion process.” In other words, after looking at a billion events on Twitter, they didn’t find a single case where, on average, one person would “infect” more than one other person.

In other words, not even viral videos and petitions spread in a manner that we would typically consider “viral.” The most popular events had elements of both “broadcast” and “viral” campaigns.

In the previous section, we found that the best way to influence sales on social networks is by encouraging audience interactions. In this section, we can expand that notion to realize that if you want to have an audience large enough to interact with itself, “viral marketing” isn’t the way to do it.

Instead, we need to focus on retention.

It’s actually fairly common for sharing activity to expand your reach by 20 percent, as previous research by the same lead author has uncovered. If you could keep every single person who saw your content, and you could expand your reach by 20 percent each time, you would only need to publish 120 pieces of content to grow your audience from one person to 3.2 billion. Since it’s actually fairly easy to expand your initial reach by 20 percent, it should be immediately clear what’s really keeping your audience from growing.

Most experienced social marketers can quite effortlessly expand their initial reach by well over 20 percent, but most of them will also lose an even larger portion of their audience with every piece of content that they publish.

If you want to grow, you need to stop thinking about producing content “worth sharing,” and start thinking about how to produce content worth coming back for.

And that’s a very different mindset.

There’s another important insight from this study: the role of influencers. If content never actually spreads in the exponential form “viral marketers” have imagined, it stands to reason that the best way to reach a large audience is to reach somebody who already has a large audience.

Of course, this is exactly what audience retention is all about: becoming an influencer yourself.

By working with influencers you not only reach a larger audience, you also spend time with people who know how to keep one, and learn lessons in the process.

3. Earned Media: Traditional vs. Social

If I asked you to name one market sector where social media definitely influences sales, I’m probably not taking a big risk in assuming that you wouldn’t say, “microlending.” Nevertheless, a study conducted by assistant professors from the University of Pittsburgh and Carnegie Mellon University, and published in the Journal of Marketing Research,  discovered exactly that.

“Earned media” has become a bit of a buzzword in the internet marketing community lately, but it’s easy to forget that earned media existed long before the internet in the form of media coverage. This study looked at both.

The researchers looked at 14 months of data about a microlending marketplace:

  • Once again, yes, they looked at actual daily sales data
  • Kiva is a nonprofit, and they didn’t use any paid media, so there were no paid campaigns to interfere with the analysis
  • Data on coverage in newspapers, magazines, television, radio, blogs, and discussion forum posts
  • “Blog” content published by traditional media organizations, like newspapers, were considered “traditional” media
  • They used Google Trends as a control measure, to make sure that they were measuring the impact of media coverage, rather than the impact of demand for information about Kiva and microfinance
  • They also controlled for the holidays and owned blog content and press releases

Again, as you can see, we’re looking at meaningful outcome metrics (sales) and we’re looking at seasoned professionals who know how to avoid conflating cause and effect.

They found that all three main types of earned media significantly impacted sales:

  • Traditional earned media events earned them about 894 new sales and 403 repeat sales
  • Each blog post earned them about 90 new sales and 63 repeat sales
  • Each community forum post earned them about 99 new sales and 48 repeat sales

After digging deeper into the data, however, they found some surprising revelations:

  • If you look at percent changes in community posting, this influenced sales 30 times more than percent changes in traditional media coverage.
  • Using the same approach, changes in blog posting frequency influenced sales 3 times more than changes in traditional media coverage.
  • Blog posts had an influence on traditional media coverage
  • Community posts had an even stronger influence on blog posts

In other words, it would be extremely foolish to assume that the traditional media coverage was more important than the community activity. In reality, sales were being primarily driven by community activity, reflecting what we learned in the first study above. Just as importantly, blog and traditional media coverage were being driven, in part, by community activity.

This study also brings with it an important insight: these discussions were taking place on forums, and most of the discussion was taking place on Kiva’s own forum.

Forums have been largely forgotten by social media marketers, even though more people visit forums than read blogs, and about 45 percent of American social media users visit a message board at least once every 24 hours. This study emphasizes just how important they are as a part of your social media strategy, and this makes sense, for two primary reasons:

  • Businesses with topically-oriented solutions are more relevant on forums, which are built around topics rather than friends and family
  • It makes much more sense to encourage interaction on a platform that you own than on a social media platform filled with distractions and algorithms you don’t control

As marketers, we can’t always wait for the data to catch up to our hunches, but we are foolish if we ignore it once it arrives, and the data is telling a fairly consistent story. Audience retention and interaction are key: reach is secondary.

Thanks for reading. If you have anything to add, we’d love to see it in the comments.

Article Banner Image

Get our best tips. Join the smartest marketers who receive our twice monthly update.

B2B Marketing Just Got Easier.

Everything you need for your B2B Content, Influencer Marketing, and Lead-Generating Research Strategies.

b2b influencer

Ready to Get Started?

I am looking for...