How did you feel about the Denver Broncos winning Super Bowl 50? Never mind, I already know, or have a pretty good idea at least. That’s because at Northwest Cadence, we built out an engine in Microsoft Azure to analyze tweets in real-time then display the results on a Power BI dashboard. This means when we were watching the game we knew exactly how people were feeling about it in the moment. In fact, we also watched the twitter sentiments spike and dip during the halftime show and commercials.
Overall, we read and analyzed over a million tweets in the week leading up to the Super Bowl, and over 5 million tweets during the game itself. (Way to go, Twitter football fans!)
If you follow our Co-founder and Strategist on Twitter you were able to see Steve Borg’s tweets during the Super Bowl. He was keeping everyone up to date on the current sentiment shown on our live Power BI dashboard. If however you’re one of the few people who doesn’t follow him on Twitter (and you really should) then this blog post will cover some of the highlights.
We decided we would analyze a few keywords for each team and the Super Bowl itself. For the Panthers we chose “Panthers” and “Keep Pounding”. The Broncos’ keywords were “Broncos” and “Denver Broncos”. Finally, for an overall sentiment of the game we looked at “Super Bowl” and “Super Bowl 50”. We then pulled in all of the tweets we could that contained the chosen keywords and analyzed the text for emotional content. It was then scored and pushed from Azure to Power BI.
What we found was that before and during the game there were about 80% more tweets containing the Panther’s keywords than the Broncos’. This did not reflect in the average sentiment though. Going into Super Bowl 50 both teams showed an average sentiment score of slightly positive. When the game started however the sentiment fluctuated with each play.
Note: In the Real Time Sentiment by Topic graph, Broncos are represented by the orange line, Panthers by blue, and Super Bowl by red.
Watching the twitter sentiment scroll across the screen while the game was on was interesting yet sometimes predictable. For example, the sentiment would invariably spike for the scoring team before returning to the mean. Gut reactions to the commercials shown however, were less predictable; some ads were not well received.
But overall, people seemed to have a positive reaction to the commercials.
Oddly enough, just after the Broncos won Super Bowl 50, the reaction toward the Super Bowl itself fell, even as the sentiment for each team stayed steady. Why would this be? (No really, I’d love to hear possible explanations!)
At any rate, next year we plan to build a more sophisticated engine for our big data analysis to gain deeper insights to questions like that.
No matter how you feel about the Broncos winning Super Bowl 50, I think you’ll agree that Steve was right when he tweeted, “Super Bowl 50 is much more fun [with] a Twitter sentiment tool built on [Microsoft Advanced Analytics] & [Microsoft Power BI].”
Who would have thought that Microsoft Advanced Analytics could be so fun?