Dr. Niels Schillewaert of Insites Consulting discussed social media research at the University of Georgia’s 30th anniversary celebration of its MMR program.  He opened with a story of his two boys building towers with toy blocks only to joyfully knock them down. This reminded Niels of the “creative destruction” of economics and of the changes affecting market research: we have to knock down towers we’ve built in the past and create new research towers. “Our clients want to be inspired,” Niels said. “The marketers who make decisions think research is boring, and we know that people do not pay attention to boring things.”

Social media is certainly not boring to marketers. Nor is it boring to researchers. According to the most recent GRIT study, 53% of researchers plan to conduct social media research, but only 11% know how to do it today. Niels offered some guidelines for the 89% who aren’t sure how to proceed. To start with, he sees two types of social media research: wild gardens vs. walled gardens.

  • Wild gardens – While people don’t respond to surveys as much as they did, they do create an enormous amount of content and post it online: status updates, blogs, photos, videos and mores. Listening and netnography are the techniques for exploring these wild gardens. “Wild gardens are about observing rather than asking questions.” Many technology companies are conducting social-media monitoring but Niels argues that “as researchers we need to claim this ground”.
  • Walled gardens – If you want to ask questions and engage freely with social-media participants, you need to cultivate “walled gardens”: short-term and long-term research communities and multi-media consumer-led ethnography projects.

Niels argues that the sheer accessibility of the wild gardens of social media lead researchers astray. Many move immediately to tracking and monitoring, without really understanding the context of what they’re monitoring. He advocates instead beginning with qualitative research. “Starting with monitoring tools first is disappointing. You need to do some qual first to understand what metrics are important to track.” Before implementing monitoring solutions, develop an analytical framework that describes the types of conversations you see in the wild gardens. Use both top-down methods and bottom-up approaches for content and text analysis.

Don’t assume automation will solve all your problems. “There’s a lot of white noise with any social media research,” Niels said. “We were doing research on Heinz ketchup. When we scraped German websites, there were a lot of Heinz’s!” Too often the data collected can be “bogus”: “There’s a lot of need for the human in sampling and analysis. Text analytics programs are limited in what they can do. You need a combination of human and machine.”

The language you will observe in social media is often not the language used by your clients or by the industry. Insites Consulting scraped 39,812 blog posts on epilepsy for UCB and uncovered the natural language used by patients, which was quite different from the clinical language of caregivers. For instance, patients described petit mal seizures as “absence” and “mind blanks”. No matter what the category, social media research can teach you what consumers tell each other, identifying hidden patterns and blind spots in the industry’s understanding. “You can find the vocabulary of the market.”

Analyzing almost 40,000 posts blurs the lines of traditional qualitative research. “Certainly it is not representative,” Niels said. “But I think it is qualitative with a quant flair.”

Again, despite the accessibility of social media data sources, developing an understanding of the data is not rapid. “Social media netnography is not a short cut – it can take 10 weeks to do a deep-dive study, analyzing, auditing, tracking, iterating.”

With social media research, is the age of the question over?  “We get answers to questions we did not ask; we get answers without asking questions.” Despite that, “we still need to ask questions – we have the data but we have to ask questions with complementary interview-based research.”

Researchers, start gardening!