On virtually every site you visit on the Internet, you’ll find social media sharing icons. Share! Pin! Tweet! Like!

Do these activities matter? Does all of that sharing create any kind of tangible benefit for the site? Instinct would say yes, but we cannot run a marketing department on instinct alone.

To answer this question, let’s put together a formal hypothesis: if sharing matters, there should be an association between the number of shares a page gets versus the number of page views that a page gets. While we cannot establish sequencing and causality, we can at least hypothesize that sharing a page should have a positive correlation to page views. Conversely, an unshared page should receive fewer page views.

It’s also important to note that we’re not examining clickthroughs on links or any other down-funnel metrics. We just want to prove or disprove a positive association between social media shares and page views.

Let’s first get our data. I’ll use data from my personal website so as to avoid revealing anything under NDA. I started by identifying the content to check. I’ll use my blog posts for the last 18 months:


From here I’ll load them into Excel, clean things up, and match them with the appropriate social shares using SHIFT’s proprietary social scanning software:


Now we do a Pearson correlation, plotting the total number of social shares on the x-axis and the total number of page views on the y-axis, then fit a trend line to the plot:

Pageviews to total social shares.jpg

We do see a relationship, a reasonably strong one, between social shares and page views. The slope of the line is measured with a term named R-squared. R-squared is .61; if you take the square root of r-squared, you get an r-value, which in this case is .781. What does that value mean? The generally accepted meaning of r values is:

+.70 or higher Very strong positive relationship
+.40 to +.69 Strong positive relationship
+.20 to +.39 Moderate positive relationship
-.19 to +.19 No or weak relationship
-.20 to -.39 Moderate negative relationship
-.40 to -.69 Strong negative relationship
-.70 or lower Very strong negative relationship

Thus, we can say there’s a very strong positive relationship between social shares and page views. We cannot say which is the driver; does excellent content that garners page views in turn drive social media shares? Or do social media shares drive page views? The data cannot answer this question for us.

To dig a little deeper, which social network shows the greatest association between social shares and page views?

Pageviews to individual networks.jpg

We see that for this data set, LinkedIn matters the most (r of .71), followed by Facebook (r of .65), then by Twitter (r of .39, a moderate positive relationship), and finally Google+ (r of .32). Across the board, there is a relationship between social shares and page views.

Social media shares matter; they simply matter unevenly. The discrepancy above means that sharing might matter more on some networks than others, or that some content resonates better with certain social audiences than others. Given that my site is a B2B-focused website and blog, it would make logical sense for strong affinity between LinkedIn and the content.

What would you do with this information? Test, of course! The first logical test would be to see if distribution matters more than content. I’d turn off my social posting for a week and see if it’s my social sharing that’s driving page views. If page views drop like a rock, then I’ve begun to establish causality. If page views remain constant, then I know it’s not my social sharing that’s driving page views, and I must develop a new hypothesis.

Alternately, if not sharing isn’t an option, I could increase the amount of sharing I do. Instead of sharing a blog post once a day, I could schedule it to be shared 2, 3, or even 4 times a day. This would test causality as well – if I saw page views increase by a proportional amount, I’d know that sharing was driving the results.

To get even more advanced, I could choose to increase sharing on the underperforming networks while leaving sharing the same on the networks that are performing well. I’d be looking for those networks to have more shares, and if I’ve established causality with page views from a previous experiment, I’d look for page views to increase commensurately.

Do the same level of data-driven analysis on your own content. Identify what content gets shared the most, which networks it is shared on, and what causes what. You’ll develop a strong understanding of what’s really working as you build your audience, create engagement, and ultimately drive business results.