Twitter predicting the 2012 US presidential election? Lessons learned from an unconscious value co-creation platform

Throughout the history of elections, political marketing services have led significant efforts aimed at predicting electoral outcomes as essential evidence to refine campaign tactics. This study develops an analytical procedure based on the Wisdom of Crowds effect and on a supervised approach of text analytics over social media content to predict electoral outcomes. Direct application of this procedure is illustrated analyzing 508,000 tweets about the 2012 US presidential election, obtaining results that consistently predicted President Barack Obama as the victor from seven weeks before the election. The study outperformed several traditional polls and similar studies employing social media to estimate potential election outcomes. This procedure offers an efficient alternative to political marketing services and political campaign staff practitioners interested in developing electoral predictions. Contributions to the field, procedural limitations, additional opportunities for knowledge creation, and research streams derived are introduced.