Havas Cracks Social Code, Develops Method To Attribute ROI

• by Steve McClellan  @mp_mcclellan, August 8, 2011

A little over a year ago, eMarketer reported the results of a survey that showed nearly 70% of U.S. Facebook users were more likely to buy a product or visit a retail outlet based on a positive referral from a Facebook friend. But according to comScore data collected just a few months later, social networks by themselves were credited with just 1% of the purchase activity by U.S. online buyers.

Havas Digital says the two stats indicate the disconnect between social network usage (Facebook alone has over 150 million users in the U.S.) and the credit (or lack thereof) the platform gets for influencing consumer purchase behavior.

One of the problems, according to Katrin Ribant, executive vice president, data platforms for Havas Digital Global, is that the industry has not experimented enough with alternatives to the so-called "last click" attribution model. This often provides online search activity with more credit than it deserves at the disadvantage of channels such as display, online video and social media for much of the country's consumer purchase behavior.

"We don't think there's a silver bullet," she said. And probably not a single methodology for measuring the ROI impact of all platforms in the digital space. But Ribant believes Havas, working with data cruncher Concentric, has reached a milestone on the road to more accurate ROI measurement across digital channels that are higher up in the purchase funnel than search.

At the core of the effort is a complex mathematical technique called "agent-based modeling" that the two companies, working with a handful of Havas clients, have applied for the first time across search, display, digital video and social media platforms. The process attempts to simulate the actions and interactions of individuals or groups of people with commonalities using computer modeling.

In one client case study, using the new technique, social media was credited with nearly 10% of the buying activity across one campaign. By comparison, social media showed no ROI impact using other methodologies.

The ABM technique itself is not new, and has been used for years for community planning, the development of complex traffic systems and other applications that attempt to predict human behavior.

Indeed, that's what previous ROI standards are lacking, said Ribant -- the ability to identify differences in individuals and consumer groups and how they react to different media. Right now, Havas is working with just a handful of clients, but hopes to have a lot more on board using the new process by year's end.

"It's a milestone, but it's still a work in progress" said Ribant.