The big data behind Obama's campaign victory
President Barack Obama's campaign for re-election rested heavily on a busy data mining operation, complete with an analytics teams five times as big as the one in 2008, mysterious code names, and a chief number cruncher. Just what this operation was doing was held close to the vest during the campaign, but Time magazine's Michael Scherer got the inside scoop.
The data mining operation was instrumental in helping the campaign raise $1 billion, target ads and model voters in battleground states, senior campaign advisors told Scherer. To boost fundraising efforts, the data crunchers started by creating a single system for gathering data from a huge variety of sources, including field workers, pollsters, fundraisers, consumer data and social media and merging it with the main Democratic voter databases in the key states. This massive database helped the campaign predict what types of voters would be persuaded by what types of appeals.
"We could [predict] people who were going to give online. We could model people who were going to give through mail. We could model volunteers," a senior adviser said. "In the end, modeling became something way bigger for us in '12 than in '08 because it made our time more efficient."
To get a clearer view of voters in swing states, the analytics operation took advantage of four sources of polling data. They were able to follow each demographic and regional group's moves in real time. They ran computer simulations of the election every night to determine Obama's chances in each battleground state, enabling the campaign to reallocate resources as needed.
- see Michael Scherer's article at Time