Wanted: Long-Term ROI Metrics

• by Erik Sass , Staff Writer @eriksass1, July 20, 2010

I don't think too many people would argue with the statement that social media is both similar to and different from other media. It has reach comparable to broadcast TV and radio, and frequency like print newspapers used to enjoy. On the other hand it is a "lean-forward" activity and obviously differs from traditional media in including a large proportion of user-generated content. And that's just the tip of the iceberg: you could probably spend a couple hours listing areas of similarities and difference.

But one of the biggest differences, in my mind, is the time scale over which social media marketing takes place -- or rather, the time scale over which it will take place. Because really, the practice is still in its infancy: by most measures social media has been around for less than ten years... but it will never go away, meaning we are just getting started. And this in turn makes me wonder if anyone is really prepared to measure the effects of social media on consumer behavior in the long term.

Social media holds challenges, but also enormous potential, for marketers who can figure out how to track consumer perceptions and engagement with brands over a much longer period than ever before.

True, there have been surveys of consumer sentiment which attempted to determine, in a fairly vague and general way, how consumer perceptions of brands and products were shaped by traditional media. These surveys could even potentially build data files for individual respondents detailing how their perceptions changed over time, provided they followed the same respondents over many years. But these surveys could never come close to comprehensively mapping the sentiment (and behavior) of every single consumer over time, let alone correlating this with specific brand messages; at best, respondents were treated as statistical data points representing broader segments of the population at large.

Social media, by contrast, offers unprecedented insights into consumer sentiment through conversation mining, and there is clear potential for drawing connections between sentiment (as well as purchase decisions, both on- and offline) and advertising. And it's here to stay, meaning clever marketers should theoretically be able to track engagement with individual consumers over years, decades, literally their whole lifetimes.

The question is, how? Advertising models built on reach and frequency still dominate media planning, but the landscape opening up before us threatens to push these models past their breaking point. Reach and frequency are hard enough to measure over the course of a typical traditional media ad campaign (or a typical online campaign, for that matter), unfolding over the course of several weeks or months. On the other hand, the fixed nature of these campaigns made it easier to determine their ROI, through short-term sales lift or some other concrete metric.

By contrast, imagine social media user Jane Doe. Jane is 18 years old in 2010, when she joins Facebook. Assuming Facebook still exists 10-20 years from now, it's possible she will have viewed thousands of ads on the site, while simultaneously engaging in conversations where she recommends and receives recommendations about brands, interacts with official company representatives for customer service issues, etc. Each one of these actions or interactions could, presumably, influence her purchase decisions way down the line. But how do you establish a causal connection between Ad X, viewed in 2012, Conversation Y, in 2013, and Purchase Z, in 2015? These events take place years apart, and occur amid a flood of other brand messages and conversations and purchases occurring in the intervening periods. It seems like a daunting, almost impossible task, but certainly worthwhile if you believe that X, Y, and Z are in fact connected (and represent a return on investment in advertising, for the money spent on Ad X).