Breaking the Social Code: Leveraging Behaviors and Attitudes on Facebook
Run in late May and early June, it targeted Facebook users in Iowa and New Hampshire with messages associated with seven potential candidates in the Republican race. The campaign had people register their attitudes by clicking the "Like" button on randomly displayed image combinations with five key Republican messages to see how specific messages resonated for candidates. Perhaps not surprisingly, the "Like rate" for ads containing anti-Obama messages ran strongest with 26%, followed by healthcare (21%), economy (18%), values (17% and national security (17%).
But the research campaign also showed how the messaging reverberates differently among the candidates. For instance, while "values" trailed other messages overall among voters, it was a leading message for Sarah Palin. On the other hand, the economy, which was a key driver for all Republicans, was not necessarily strong when associated with Palin. And while the economy is an important force on everyone's minds, this research suggested that healthcare messaging was even more impactful.
This kind of campaign is possible because of the tools Washington Post had developed in-house over the years for serving ads. "The technology and the serving method was developed at Washington Post Company," says Laura O'Shaughnessy, General Manager, SocialCode. "We thought this could be used to help get brand marketers onto Facebook and so we applied it to the Facebook Marketplace." Using the social network's beta API program, SocialCode can put out their own tools to serve ad units that run in the Facebook Marketplace slots on the network's pages. SocialCode acts as an agency within WashPo to advise and execute client ad strategies here. "At a high level we help brands survey their marketing strategy - whatever their brand's goals are - and help them design it for the Facebook platform," she says.
The idea is to leverage the social network data to understand who is proving most receptive to the ads being served. "We basically cast a wide net and see which profiles are responding," she says. "We build up data on those profiles to see what messaging works best." They are looking at the available non-PII data Facebook offers, from geographic location to age, sex, relationship status and interest profile to create segments and test messaging against them. "What is most important is what drives the brand's success," she says. "If you are a B2B brand and you are looking for C-level executives, then that defines our strategy and we develop a method to target those users." Or you can use the tools to help identify who you are really looking for. "For a CPG brand, they wanted to test which targets would respond. So we developed six or seven target categories with certain themes, like a snacking or dieting theme, and we were able to effectively drive qualified customers that way."
Ultimately, this kind of social media tool works best when there is a kind of virtuous cycle of research and targeting, which begets more results to inform later campaigns. Over time the SocialCode model comes to understand who a client's audience is on Facebook and how they interact with certain messaging and offers. "There is a data orientation," she says. "To build up profiles we will cast out the net when we first start and test a brand's potential target base. Then we quickly find out which profiles are responding to which messages and hone in on those targets. Then we learn more about who these people are. We can work with brands over time to build out those data learnings. We start with one campaign and then with a second we use the learnings from the first engagement. So we come to know more about the most valuable customers to that brand, the ones who sign up for things and make purchases."
O'Shaughnessy says that SocialCode works best with clients who are committed to the data and who want to leverage the dynamics of the social space in campaigns. As the same time the richer profiles that accrue over time can also inform a wider range of display buying and messaging off of the social network. A "like" may be a simple response metric -- but when exercised on the social network it is tied to a wealth of data about that user that traditionally has been absent from the equally simple "clickthrough."