The study, published in the Journal of Marketing Research, compared social media data and metrics with traditional brand tracking surveys and found that social media data can serve as an indicator of customer satisfaction measures subsequently obtained from brand tracking surveys. The authors developed a model to use this data as a measure of brand health, as well as diagnostic metrics of brand attributes.

The model outlined in the paper allows for the idea that opinions posted by customers online may be subject to a number of factors that might mask their true underlying opinion toward the brand. These include “venue choice” – i.e. the type of online forum they choose – and the difference between opinion toward the brand and toward specific brand attributes.

“For example, consider a customer who loves their Apple device but turns to social media to express concerns about price,” say authors, professors David Schweidel and Wendy Moe. “Most text analysis methods would code the comment as negative in sentiment. However, the poster’s overall opinion toward the Apple brand may still be positive.”