The Obama campaign and data
The Political Marketing Group asked me to do an article for their March newsletter. There might be some wider interest in it, so I thought I would blog it here also:
Barack Obama was re-elected for a number of reasons including the quality of the candidates, their policies and their records. But one of the reasons was also the quality of their campaigns, and the Obama’s campaign use of big data to bring a new level of sophistication to political marketing.
The days of campaigns being about getting the best coverage on the 6 pm news every night are well and truly over. By 6 pm, many people already know how the campaign has gone for candidates. The buzz on Twitter has often made it clear how the day’s happenings will be reported.
The Obama campaign used data to divide voters into three categories. Those who were not worth pursuing who were left alone, those who were moderate supporters who might donate if asked and those who were strong supporters who might become activists.
They used research to survey millions of voters so they could sort them into the three categories. The research was a mixture of postal surveys, phoning and visiting. This data was supplemented by social media data and advertising.
An interactive Facebook video would depict to individuals how the President’s policies would help them and their friends. But the real purpose of the video was to get their permission to siphon off data about all their friends so they could be matched to their state voting records.
Having collected so much data on voters, the Obama campaign then used it to personalise online advertisements and messages. Their data told them the most appealing celebrity for middle-aged women on the East Coast was Sarah-Jessica Parker so they used her to appeal for their votes.
Having done badly in 2010 due to low turnout, Obama’s campaign focused on identifying those who voted for him in 2008 and ensuring they voted again. Every voter in the country was assigned two scores. One being their likelihood to vote and one being the probability they would vote for Obama. They then calculated for each individual precinct who were the likely people who voted Obama in 2008 and worked on making sure they voted in 2012.
They also used data to test their messages. Up to ten different varieties of an e-mail solicitation would be sent out to a test group. The communication that achieved the highest response or donation rate was then used for the entire population. Almost every single message and communication was tested scientifically. It didn’t matter so much what the creative director though of the communication. What mattered was measuring what impact it had.
Advertising has been the traditional channel for persuasion in political campaigns. It still remains an important element, especially as they can impact media coverage also. But the lessons from the 2012 United States presidential campaign are that advertising alone is most definitely no longer enough. Data, social media and electronic communications are the weapons now used in a 21st century campaign, and political parties and candidates that fail to use them will struggle to achieve the result they want.