Bloom Correctly Predict Eurovision Song Contest Result

At Bloom, we love nothing more than putting academic theory into practice and using this to help businesses connect with the latest changes in technology.  Having spent the past few months working closely with the University of Reading’s Centre of Mathematics for Human Behaviour to develop a next generation social media listening tool, the Eurovision Song Contest provided a great opportunity (in partnership with Datasift) to test our research in a real-time environment before rolling it out to a commercial market later this year.

In the run-up to Eurovision, Sweden’s entry (‘Euphoria’, performed by Loreen) had been number 1 in Sweden and Finland and had top 5 positions in 4 other European countries. The bookies expected the song to do well, but quite how well was anyone’s guess.

However, Bloom’s social listening tool had also predicted a Swedish victory. It had been hard at work listening to conversations around Eurovision on Thursday’s Semi Final evening, tracking over 320,000 tweets between 8pm and 10pm. At its peak, just after 9:12pm, an amazing 4,500 tweets per minute had been received.

 

During the Swedish performance, a magnificent 40% of the conversation was related to the Swedish entry and they received the biggest spike when the results were announced.

So, how did we know whether people were saying good or bad things about the Swedish entry? After all, the Netherlands and Portugal also had spikes in conversation, but they didn’t make it through to the final.

 

Bloom’s tool was able to aggregate the sentiment of the conversation in real time. As the graphs above show, the spike in conversation about Sweden coincides with a massive jump in Sentiment. People were talking about Sweden – and, most importantly, they were saying more good things than they were bad. The Sentiment about Sweden continued to stay high during the rest of the programme, whilst Portugal and the Netherlands received no spikes in sentiment. It was Sweden’s spike in perception, as well as overall share of conversation, that we used to predict their victory.

Flash forward to the Eurovision song contest final two days later. Whilst the rest of the country was settling down with a stiff drink or three to watch Englebert Humperdink belt out ‘Love Will Set You Free’, our analysis team were sat biting their nails, wondering if Bloom’s tool had been correct in predicting Sweden’s victory.

The answer was a resounding yes – Sweden’s Loreen achieved the second highest score in the contest’s 56 year history, picking up more votes than Russia and Serbia during tactical voting and achieving a bigger share of the ‘non-political’ vote.

So what does this tell us about Bloom’s social media tool? Well, it can successfully listen to social media conversations. It can help brands to understand which individuals are influential in a particular conversation in real time and it can allow brands to develop real time strategies to engage and foster communities online.

We’re really excited about the possibilities.