Click here to learn more about author Mark Green.
If there’s anything that shows that social media analytics are ready for prime time, it’s the Brexit vote, Britain’s decision to exit the European Union.
The Brexit debate was a rollercoaster on TV and in homes around Great Britain. But with the right insight, it was actually possible to use social data to predict the outcome of the vote. In fact, our analysis of Twitter, which tracked the weeks of surges and dips in opinion, indicated the day before the vote that the Leave camp would win. And it did.
Not all social media prognosticators came up with the same result. That’s natural. The reliability of any analytical tool comes down to methodology. Being able to listen and understand what more than 1.5 billion people are doing on social media is potent. But the real power comes from being able to reliably pull actionable insights out of that noise.
We do that by tracking all the social engagement data points across different social media services and identifying affinity points between this activity around brands or topics. Rather than studying one topic in isolation, our approach is to understand how one topic relates to many topics and over time, the shifts in these relationships. At the same time, we study the relationships in terms of intensity and frequency, or how often many people are engaged in a topic. Examining this combination of factors provides insights that aren’t obvious from just tracking the sheer number of tweets or Facebook comments about something.
Effective analytics isn’t about listening to celebrities or the loudest voices, it’s about sound science. The media will always report on the most well-known or provocative voices. But good social media analytics doesn’t focus on the loudest voice. It combines the voices of the individuals into a signal of the masses and it provides the ability to track those signals 24/7, creating a reliable reading of trend lines over time.
We’re reaching a pivot point where social media will have a big role to play in explaining voter sentiment. A much bigger portion of the population is now on social media–more than two-thirds of the people in the U.S. This is an audience that can provide a better read on opinions than some types of polling. Polls require random participation and extrapolation, which is becoming harder to get. Social media enables you to know the opinions and preferences of people through the best indicator possible — what they’re actually saying and doing. Not just at one point in time, but over days, months, years.
Social media analytics is far from perfect. Not all topics are discussed on social media with the same level of interest, and that can make it hard to gauge relative interest or compare topics. And though the use of social media continues to climb, not everyone is tweeting or posting on Facebook or Instagram. What we have right now is a subset of the population, though it’s the largest subset and most meaningful sample of data from the population ever available.
Some pundits argue that social media analysis got it wrong with Brexit. It is true. Some methodologies did not work well. But social media did not get it wrong. When you have the right methodology, paying attention to the right signals, social media can give you insight into what is happening versus traditional approaches which have become more expensive and struggle to remain accurate.