Three ways to turn online discussions into insight

By Jeffrey Henning

Ray Poynter, managing director of The Future Place and organizer of The NewMR Virtual Festival, kicked off the festival with a presentation on “How can we turn online discussions into insight?” He suggested three strategies: netnography, discussion creation and mass techniques.

• Netnography – To see where academia meets real practice in the real world, check out Robert Kozinets book Netnography. Kozinets outlines the practical steps that you can follow. To him, netnography always involves observation and participation. It is not enough to simply monitor what others are saying online: you need to participate in the conversations and echo back your analysis to other participants to make sure you really understand what you’ve heard (what Kozinets calls member checking). Once you’ve collected a corpus of discussions to analyze, there are a wide range of techniques you can use: discourse analysis, conversation analysis, hermeneutics, grounded theory, semiotics, NLP, word counting and mathematical models.
• Discussion Creation – If the Australian tax office can have a conversation with their “customers”, Ray argues, then all organizations can have conversations with their stakeholders: their customers, employees, partners. Look to social media to create conversations, and look to novel services for new opportunities to create conversations (for instance, Foursquare). “Market researchers wouldn’t want to create a conversation and feed it into Marketing,” Ray argues, “but if Marketing creates a conversation then that should feed into Market Research.”
• Mass Techniques – The third approach is mining vast quantities of data, reporting on trends. How many people mention a brand name? A movie star’s name? What are the trending topics in a community or about a brand? Mass techniques leverage content to see changes in its volume, to search for trends, to perform sentiment analysis and to identify influence. Some of these mass techniques are quite inaccurate; for instance, FreshMinds looked at how bad most sentiment analysis software is. Others have looked to identify the role of influence on social media conversations.

Social media research ethics are still being discussed and parallels are being sought. For instance, “looking at the ethics of mystery shopping – what are the rules there and are they applicable?” Preserving anonymity is really difficult, when someone else can use a search engine to identity the commenter behind a verbatim quote. Already firms like Google, Facebook and Nielsen have run into privacy debates.

The final issue in turning online discussions into insights is confidence. “In quantitative research, confidence is measured statistically: 90% certain with a plus or minus 3%,” Ray said. “Now we have to say: ‘Some people believe this, lots of people believe that, nearly everyone we were able to find believes in these.’ To achieve confidence, some sort of member checking is required, going back to some of the people to see to verify that you have correctly understood what they meant.

Ray summed up by saying, “This is going to be BIG! But there will be disasters!” He advised that “software is better at finding corpora than analyzing them” and that researchers need to know what model they are applying to their data and need to educate clients about the confidence they can place on the findings you report.

Republished with permission from the Voice of Vovici blog.