Getting More-Reliable “Intelligence” from Social Media Data
Marketers are turning to social media data to understand consumer sentiment, but do current metrics offer a reliable picture of how consumers feel about their brands? Not according to a 2012 study, “Social Media Intelligence: Measuring Brand Sentiment from Online Conversations.”
“Firms are relying on simplified measures like the total volume of posted comments or average sentiment expressed in venues such as Facebook, Twitter, blogs, and forums,” says coauthor Wendy Moe of University of Maryland. “The problem is that each of these is a very different environment that can lead to very different posting behaviors.”
Moe, who will speak at MSI’s December conference on “Social Media and Social Networks,” is an expert on social media listening, web analytics, and forecasting. Her research has focused on developing methods and models that can help marketers understand and use social media data more effectively.
In the 2012 study, Moe, with David Schweidel of Emory University and a team from the social media consultancy Converseon, developed a statistical model that adjusted for various factors—such as venue and posting dynamics—that can influence posted opinions. Importantly, their model allowed them to isolate a measure of underlying brand sentiment from how customers felt about individual product attributes within the brand portfolio.
They tested the model using two datasets focused on a single enterprise software brand. The first was provided by Converseon, which scoured the Web from June 2009 to August 2010 to collect comments relating to the brand. Converseon’s analysts determined the focus of each post, distinguishing between 140 unique products in the brand’s portfolio and 59 brand attributes.
The second dataset was collected the old-fashioned way: A telephone survey of 1,055 registered customers given on a rolling basis over 10 months coinciding with Converseon’s data collection period.
Moe and her coauthors found almost no correlation between the telephone survey results and a blanket measure of average online sentiment. When they compared the telephone survey’s results with their model’s adjusted measure of online brand sentiment, however, the correlation was approximately .80, which is quite a strong relationship.
Their findings demonstrate the potential for social media to be incorporated into the brand’s research activities, but only if marketers use the right measurement tools. “Marketers must cast a wide net across venues when monitoring social media. To understand underlying brand sentiment, they need models that account for differences across those venues and customers. Aggregate measures may offer ‘easy’ answers, but they don’t provide a reliable understanding of how customers feel about your brand,” says Moe.
“Social Media Intelligence: Measuring Brand Sentiment from Online Conversations”
David A. Schweidel, Wendy W. Moe, and Chris Boudreaux (2012)
Social Media Intelligence
Wendy Moe (2013) [Video]
3 WAYS to GET CONNECTED
Employees of MSI Member Companies enjoy the benefits of complete online access to content, member conferences and networking with the MSI community.
Qualified academics benefit from a relationship with MSI through access to msi.org, conferences and research opportunities.
The public is invited to enjoy partial access to msi.org content, a free e-newsletter, selected reports and more.