Social Media ROI Revisited: A Look at Correlation

In the immediate aftermath of the now-epic Old Spice Guy YouTube response campaign, Facebook fan interaction spiked by 800%, the brand’s Twitter following grew by 2700%, 1.4 billion impressions were generated, and more importantly, product sales surged 27% over six months to establish Old Spice as the men’s body wash category leader.

Were all those sales directly trackable back to people who viewed the videos or shared them with their Facebook friends? Probably not. Rather, the argument being made is that the surge in online conversations, sharing, and all-around buzz was strongly correlated to the boost in sales.

Calculating return on investment (ROI) for social media activities isn’t always as straightforward as I outlined in my last post, exploring examples of direct ROI. Sometimes, as with the Old Spic Guy campaign, it’s just not possible to directly track the financial impact of each and every click or conversation. When direct ROI isn’t an option, it’s time to consider the other three models of social media ROI I outlined in my first post on the subject: Correlated, Relative, and Proxy.

For this post, let’s look at correlated ROI. Correlation may not give you a definitive  measure like direct ROI will – invest $1, get $1.50 back, and so on – but it can be very useful as a general indicator that tells you if all that online conversation and activity you’re driving is generally good, bad, or irrelevant to the bottom line.

Old Spice is something of an extreme example – it did win just about every award on the planet – but the same general principle can be applied for your social media campaigns, promotions, and activities as well.

Correlated ROI Defined

As I said in the first post:

Correlated ROI is a function of tracking measurable social media activities over a given time – a focused blogging effort, launch of a new community site, a Facebook campaign, etc. – and comparing it to the performance of key business or marketing metrics, such as sales volume or customer service calls, over the same period. You’re looking for a relationship between the two data sets, with an ideal of being to identify that for a given investment in social media efforts, or a given level of activity, you’ll get a corresponding impact on a key business metric like sales.

In the face of an overwhelming online campaign like Old Spice, making the correlation argument seems like a no brainer, though there was (and likely still is) a bunch of debate on that point. But how does this play out for something not quite so obvious, say for the amount of overall investment in social media channels by a big company?

Speaking at SXSWi earlier this year, David Witt, the Global Head of Social Engagement and Brand Public Relations at Hershey’s and formerly at General Mills, spoke of correlation while at the latter company:

Witt noted that he has repeatedly charted online conversations and actual sales and the lines move in virtually perfect symmetry. The only stronger correlation was being on shelf. In essence, Witt said, we’re now showing strong correlations more than definitive ROI, but the numbers are very impressive. Eventually we’ll have to show hard numbers, he said, but today it’s a bit difficult to do so definitively.

So for General Mills, a higher volume of conversation strongly correlated to actual sales – the direct line from buzz to sales might not be there, but correlation at least suggests that driving that conversation is well worth the company’s time.

For another example of correlation in action, back in 2009 the Altimeter Group and Wetpaint put out a popular study that made the claim that social media engagement – an aggregate metric they came up with to represent how active a brand is on various social media channels – and financial performance were positively correlated. They analyzed 100 top global brands, grouped them into four categories from “wallflowers” to “mavens” and compared them to recent financial performance. They found the Mavens significantly outperformed the Wallflowers and other groups when comparing revenue growth, net margin growth, and so on.

The pitfalls of assuming causation

Now even the study authors note that this correlation does not necessarily mean higher social media engagement caused better financial performance, simply that the correlation exists and is worth exploring. The issue is a nice example of the old caveat that correlation does not imply causation:  just because two sets of numbers seem to move together does not necessarily mean one is causing the other to move. When you identify a social media metric that appears to be correlated with a business metric, don’t just assume that if you do more of it you’ll get better results. That’s why I noted above that correlated ROI is best for getting a general sense, and setting you up to explore the relationship further.

For example: That spike in sales that appears to be nicely correlated to a sudden rise in Facebook activity might in fact have been caused by a strong coupon code offered through a related email marketing campaign, and a bunch of excited people just happened to pile on to your Facebook page to talk about it. The lesson? Don’t leap to conclusions, take the time to explore the correlation to better understand in real world terms what the nature of the relationship may be.

Case Studies & Examples

Beyond General Mills mentioned above, here are some other examples of correlated ROI for social media:

Additional Reading

Stay tuned for additional posts in this series, where we’ll explore in more detail both relative and proxy ROI and some additional practical tips. Either subscribe to this blog or follow Ignite Social Media or Kevin Briody on Twitter for the latest posts when they are available.

Related Posts