If you have been following this series of events then you know that for 3 weeks running Biz360 used social media measurement to successfully make Amercian Idol contestant predictions.
Yesterday at 3:00 pm when we needed to make our finale prediction announcement, there was less than a 1% difference between the final contestants, Adam and Kris. Many were saying it was too close to call. But we had promised a prediction and a number of colleagues and clients were making inquiries – we didn’t want to let them down. After all, the show can’t announce a draw now can they?
While the difference between the two contestants was small enough that an accurate statistical prediction was impossible due to the limitations of margin of error – we went ahead and predicted Adam Lambert as the winner based on the data we had at the time.
What made this week’s prediction different than those in previous weeks?
This week was particularly challenging because not only was it an impossibly close margin, but 40% of the source coverage being measured came from Microblogs (mostly Twitter). The week before for example, consisted of closer to 20% of Twitter-type coverage. Microblogs are about as real time as it gets, which means it is a dynamic moving target, evolving as public opinion takes shape.
We started out this exploration by asking the question: Can social media predict American Idol? At this point if we look back at yesterday’s positive and neutral coverage, Kris did wind up slightly ahead of Adam. Adam landed at a share of 49.34% while Kris was at 50.66%. Overall, we believe we have demonstrated that it absolutely can.
As a side note, a number of people were requesting a Dancing With The Stars finale prediction also. Yesterday before the finale show, we made a correct prediction analysis for the twitter community using our Community Insights solution. Even though Shawn Johnson was the expected underdog, her social media coverage lead us to predict her to win.
American Idol was a fun vehicle to prove the concept — but imagine the applications for companies. Throughout this exploration, people continue to be amazed with the idea that we can take thousands and thousands of relevant aggregated posts and conversations, and very quickly consolidate them into real, usable data. The time period between the performance show and results show is only 24 hours. Also, the people making these posts are handraisers, volunteering their unsolicited feedback. So the quality and bias of the consumer opinions is minimal when compared to traditional research methods of gathering feedback. Our clients are benefitting from this streamlined path to their customers by using Biz360 solutions like Community Insights and Opinion Insights to listen.