You probably wouldn’t expect the State Department to spend taxpayer money buying Facebook fans, but in 2011 and 2012, it did, spending $630,000 to “increase engagement with foreign audiences.” Superficially, the campaigns succeeded, and the number of fans increased from about 100,000 people to over 2 million.
Of course, that’s a ridiculous strategy!
Everyone knows that dropping cash on fans won’t increase engagement since half of them are probably bots. Shares, likes, and comments from real people who care, are the currency of the social web.
Unsurprisingly, only 2% of fans interacted with the State Department's posts according to an inspector general's report in 2013, which led to the swift termination of those campaigns.
It’s not just the State Department, though.
Social share are so desirable in fact, that a Google search for “how to improve social shares,” returned 725 million results. That’s 200 million more searches than “how to make money.”
Seeking the Holy Grail of Shareability
We’ve got a mystery on our hands. In an industry obsessed with measuring and quantifying results, we still don’t know how the magic happens. Sure there are hacks, but in fact, the seemingly straightforward act of sharing is pregnant with meaning, representing how people define themselves, how they network, and even how they experience fulfillment.
And while no formula can reliably forecast something as fickle and complex as human nature, in my office at Oz, where we're obsessed with predicting what works, we did notice that people tended to share well-reported, well-researched, credible, articles more often. Furthermore, we knew that Google placed a premium on credibility, pushing the most reliable sources to the top of the search results and that Moz had developed a Domain Authority (DA) score, which measures the trustworthiness of a site. While other studies have shown positive correlations between social sharing and organic search ranking, and ample clues indicate that linking to high DA sites increases search ranking, nobody had looked at the correlation between outbound link DA and social sharing.
Accordingly, we analyzed over 90 articles from three blogs: Contently, Moz, and Buffer for a connection between outlink DA and social sharing. While our small study confirmed the importance of DA and good reporting, it also hinted at a complex landscape where actual authority is judged differently depending on your niche, and where credibility built over time, reinforced by DA, and by subtle clues woven into the text.
Part 1: Why do people SAY they share on social media?
Nobody specifically shares an article because it is sourced from publications with high domain authorities. In fact, most Internet users probably don’t know that Moz’s DA score uses 14 factors to track the strength and reliability of a website over time.
Even so, people do admit to sharing for highly personal reasons related to status.
The New York Times Consumer Insight Group conducted a study using in-person interviews, a weeklong panel, and a survey of over 2,500 online sharers for a report called The Psychology of Sharing. The results revealed that people shared for five main reasons:
- To bring valuable and entertaining content to others: 94% of those surveyed said they carefully considered how the information they share will be useful to the recipients
- To define themselves: A participant representing this group said, “I try to share only information that will reinforce the image I’d like to present.”
- To network and develop relationships: 73% share because it helps them feel involved and connected with like minded people
- To feel fulfilled: One participant representing this group said, “ I enjoy getting comments that I sent great information [that is] so helpful. It makes me feel valuable.”
- To support a brand or cause: 84% of people share because it’s a way to support things they care about.
This is deep stuff, entwining sharing with personal meaning, prestige, and identity, “Sharing has a lot to do with the ego. If I can teach and delight you, I’ve raised my status in your eyes,” Lavall Chichester, Head of Search at Lowe Profero told us in an interview. “Readers want to learn, or to be delighted, and they don’t have time for garbage,” said Chichester, who also compared building credibility to building muscle over time. Unlike muscle, though, “credibility can be transferred to readers. If I find something and share it, I have the ability to delight and teach. The more I do this, the more people look to me as a source of reliable and delightful content.”
If that kind of innocuous egocentricity really does drive sharing, publishers and writers should create work that fosters connection amongst people, making sharers look smart, useful, and informed. It would follow then, that those sharers would be more likely to share high-quality articles replete with high DA outlinks.
Part 2: Methodology
Using Screaming Frog, we scraped approximately 90 posts published across three blogs to test if people were more likely to share trustworthy articles.
- We only chose articles written in the last few months to avoid shares accumulated over time.
- We selected text-based posts of at least 800-words, focusing on the ones that informed or taught readers, rather than product news or announcements, which tend to get fewer shares, or infographics, which tend to get more shares.
- We looked at the outbound links for each article, and using Moz’s DA score came up with an average DA score for the outbound links in each post. We did not include links from the comments, and we eliminated links to or from the author’s profile.
- We excluded all posts by Rand Fishkin, ‘cause he’s too popular, and we thought his popularity might skew results.
- We plotted the average DA score against the number of social shares that article received on Facebook, Twitter, LinkedIn, and G+.
“But,” you may object, “links are hidden, and a reader can’t even tell if a link goes to an article with a high DA score!” Oh, but they usually can. If an author is about to cite an article from in important publication, she’ll likely say something like, “according to the New York Times,” or “researchers from MIT,” or “a recent Pew Research poll.” Along with specific details, quotes from experts, and statistics, these are some of the subtle clues within the text that convey credibility and which may lead some readers to value and share your article.
Results #1: A positive correlation
Our study showed that Contently and Buffer readers were more likely to share articles with a high average DA, “It’s not surprising,” Joe Lazauskas, Editor-in-Chief of Contently, told us in an interview. “What we really aim to do is to practice really good research and reporting in the content we publish. It makes sense that our better pieces and our best pieces include a lot of very good research and reporting, which means they’re linking to highly credible sources that have a high domain authority.”
Links to high DA articles show that publications are doing their home homework and are offering the reader valuable information.
Buffer’s Kevan Lee added that the high share-rate could indirectly result from using the Skyscraper Technique, “if other, influential blogs have written about the topic before, it's perhaps a topic that we could expect to gain significant social shares when we write about it as well.”
However, while the correlation to social shares offers edifying evidence for well-reported, quality-content, don’t be fooled into thinking it’s magic. You can’t just plug in high DA links and expect people to share your article. Not going to work.It’s correlation, not causation.
Nevertheless, the evidence does reveal the potential importance of using links as a student would use references, to prove something is credible, that other people stand behind it. Perhaps, Lee says there’s an “element of social proof to an article when you're able to reference blogs and influencers that others already know about,” that gives people an extra nudge.
Lee’s point about social proof is crucial even though some bloggers only mention where a statistic comes from, failing to include the link, perhaps fearing Google will punish them for too many outlinks. (In reality, Google penalizes only those who egregiously link out, for clear link-trading purposes.)But of course, social proof is transmitted through more than just hyperlinks.
"It’s been said that when people are free to do as they please, they usually imitate one another," wrote Jonah Berger in his book, Contagious: Why Things Catch On. "We look to others for information about what is right or good to do in a given situation, and this social proof shapes everything from the products we buy to the candidates we vote for." And of course, social proof influences the articles we share.
For his part, Lazauskas stands behind proper citations, “We link to the original source,” he said. “We don’t link to some anonymous marketing blog that rounds up a bunch of stats; we find the original report. I think that’s important, and it affects the [average DA] as well. I think a lot of people are kind of lazy about that.” That strategy certainly seems to be working for Contently whose articles in our study got an average 1,378 shares per article across social networks. Wow.
Results #2: A negative correlation!
Moz is an industry superstar; we turn to their blog and products to teach us, to break news, to tell us what’s important, and even coin terms like Domain Authority and define industry standards. So how do we explain that on Moz, the most frequently shared posts had a lower total Domain Authority?
“Maybe, it’s the Rand Fishkin Effect,” said Chichester, who speculated that perhaps“stripping Rand’s posts out caused this. I think Rand’s content gets a ton more shares that anything else so that may have skewed the results.”
That’s possible, and we’re looking forward to expanding our testing and trying various control scenarios, but we’re inclined to believe that this divergence results from content type and credibility in context. Moz is full of what Rand Fishkin calls 10X content: it’s very long, very detailed, and it has a lot of outlinks.
As we analyzed the links in these stories, we noticed that a lot of them were to lower authority sites within the SEO community—links to tools, studies, agencies and blogs that were extremely helpful, but because they were so niche, did not have the highest DA.
Because Moz is a niche company focusing on SEO, largely for an SEO-focused community, they don’t reference the NYT or Washington Post as often, but they do cite the best SEO tools and thought leaders in SEO community. Lee, who was also a bit stumped by negative correlation had a similar take, “One thing I've noticed is that a lot of articles I enjoy from [Moz] are original, independent case studies or tutorials that touch on things many others haven't yet. By their very nature, those posts might not have a chance to link out to many places - and the posts themselves are so great, I'd not be surprised at all to see the social shares go crazy.”
It’s counterintuitive, but this negative correlation actually affirms our hypothesis that it’s the quality and credibility of content that gets people to share.
Authority is only one contributing factor to credibility.
In their excellent book Made to Stick: Why Some Ideas Survive and Others Die, Chip and Dan Heath explain, “It’s the honesty and trustworthiness of our sources, not their status, that allows them to act as authorities. Sometimes antiauthorites are even better than authorities.”
We found that in these 10X articles, the link ratio was much higher, with many more outlinks per hundred words than there were on the other sites, but not all those links had a high DA. In fact, most of them didn’t.
These highly shared Moz articles are rife with quality research and reporting; there’s a lot of illustration and technical pointers on how-to, but the outlinks point mostly to tools, agencies, and blogs that don’t have the highest DA. They are, nevertheless extremely useful and highly credible.
We share what we can trust
It turns out average outlink DA is a surprising way to determine how reliable a particular article might be, especially when content appears on generalist sites. However, our study showed that DA is just one measure of actual authority and that being credible is a larger phenomenon.
On generalist sites, people are more likely to share articles with a higher average outlink DA. Clues within the text, including hyperlinks, statistics, quotes, and references to well-known institutions, all transmit credibility and guarantee that the sharer won't look bad.
In the case of specialized niche sites, the most reliable sources might not have a high DA, yet they may be extremely reliable and authoritative, having built their credibility through original research over time. A high DA is a measure of status, but it turns out being credible on the Internet, being trustworthy, is far more complex and variable than status alone.