The true value of social media is not in providing a high-tech platform for people to socialize on, but the data that can be derived by an analysis of what people are saying, doing, watching and listening to. Sure, social media titans such as Facebook and Twitter are working hard to monetize their platforms with targeted advertisements and other strategies, but others in business and academia are taking advantage of social media to derive interesting and potentially valuable information about almost any topic that circulates widely on social media.
The Bank of England decided to experiment with social media analytics during the referendum for Scottish independence late last year. In a recent blog, BoE analysts David Bradnum, Christopher Lovell, Pedro Santos and Nick Vaughan describe their efforts to use the Twitter micro-blogging platform as a predictive tool to give the central bank insight into the possibility of a bank run if it looked like Scotland was voting for independence from the UK.
Of course, it didn’t turn out that way, as the vote went against Scottish independence pretty early on. This meant there was little to no “twitter chatter” on topics relating to a bank run, but the BoE analysts still consider the experiment a success given how much they learned and the massive potential of social media analytics.
The BoE social media analytics method
The BoE team set up software to monitor traffic on the Twitter platform in the days and hours ahead of the vote, looking for tweets including terms or phrases that could have indicated depositors were considering withdrawing money from Scottish banks. The basic goal was to monitor the Twitter traffic matching these search terms as an “early warning system”.
However, given the referendum ended up being clear cut against independence relatively early on, there was very little Twitter chat about Scottish bank withdrawals and obviously no early warning of a bank run was triggered.
The Minnesota Vikings anomaly explained
The experiment was very useful, however, as a teaching tool. One interesting anomaly was a spike in a key term the early hours of Monday, September 15th. However, it turned out the spike was related to American football rather than the Scottish election.
It turned out that there had been a series of tweets and retweets involving the NFL’s Minnesota Vikings. These tweets ended up being noted by the BoE softweare because they included the term “run” and the abbreviation “RBs”. However, the reference in this case was to “running backs” and not the Royal Bank of Scotland. The team then changed the search parameters so they would not include that specific pattern of upper and lower case letters.