After President Obama was re-elected in 2012, it emerged that his campaign had successfully used big data analytics in ways no other political campaign had done before. The team of over 100 people started with the core data of the voters’ roll, modelled how they could understand it and then conducted realtime experiments to see how their actions affected individual voters.
As a result, Obama’s campaign manager was able to carefully measure the mood of undecided voters and target them close to election day, winning key states in the fight.
The initial size of the data was quite modest – 10Tb – but quickly exploded as the analytics team experimented with it. Afterwards, many observers were surprised to find that the team had used a SQL database and a set of statistical analysis tools to perform their magic.
“The Obama campaign of 2012 showed that big data analytics is all about the real insight you can generate and not necessarily the fancy tools you use,” says Anton van Heerden, GM at Altech ISIS.
“Many organisations today are struggling with the inflow of terabytes, if not petabytes, of information into their data centres, and how they can make sense of it all. But the lesson of the US Presidential election is that having a focused team, careful modelling and constant experimentation with the data until you get it right is what counts.”
Van Heerden says the other factor was organisational.
“Obama’s campaign manager flattened the reporting and organisation of the teams so that they could work together more easily. And instead of hiring more engineers, they hired more analysts and gave them an easy-to-use analytical environment in which to play. Instead of requiring people with experience in complex enterprise tools, they could hire analysts with the ability to run simple SQL queries.”
Altech ISIS looks after the data warehousing and analytics for telecoms companies where analytics plays a major role in reducing churn rates and improving customer service. Telcos are no stranger to big data but even they are searching for better ways to get insight from not just their call records but the masses of other inputs into the business environment.
Van Heerden says there are some pitfalls to avoid when confronted with terabytes of information that could be mined for any insight that leads to more revenue or lower costs.
“A human, complete with biases and prejudices, normally makes the call about what to try and correlate and what not to,” he observes.
“Instead, companies should really be trying to let the data speak for itself. But that having being said, the temptation that creeps in when the data drives the business agenda is for companies who are not getting results to simply go out and get more data. You need to keep a tight control of what data you decide to analyse.”
He adds that outsourcing to a specialist company like Altech ISIS not only improves efficiencies in terms of the management of the data, analyses also improve when an independent expert is working with the information.
“Just like President Obama’s team of political experts hired a team of analytics specialists, companies get the most value when a professional whose focus is on extracting the value of the data is employed. It is his job to find the correlations, and the company can then use this to create actionable insights, just like the US President did.”
