Skip to content

From Capitol Hill to Westminster – how big data is transforming political campaigning

The upcoming General Election has been described as one of the most unpredictable in living memory. With no clear favourite, everything is up for grabs – or so it seems.

Throughout the halls of Westminster, a growing number of political parties are putting their faith in emerging technologies, believing data analytics will help them gain an edge over rivals, defend vulnerable seats, and strike a closer cord with the electorate.

It might sound like a move straight from the Silicon Valley playbook, but it’s one with proven pedigree. Obama’s last two presidential campaigns made widespread use of big data analytics to gauge sentiment among the voting public in order to adapt strategy while on the campaign trail.

So impressed were Labour and the Tories that they have both hired senior Obama advisors to work alongside their campaign teams, hoping to ape the success of the Democrats in securing key voters and states.

The 2008 US election is considered to be a watershed moment for political campaigning. As one former Obama campaigner famously said: “We stopped thinking in terms of ‘soccer moms’ and started thinking instead about ‘Mary Smith at 37 Pivot Street.’”

Data was gathered and compiled to create a complex profile of localities and individual voters. Myriad sources such as private polling, membership logs and canvassing were logged and fed back to party HQ, combined with location and voter ID information from the electoral register.

Social media played a massive role. While the President’s significant Twitter following dwarfed that of challenger Mitt Romney’s, the Obama camp also spent $47m on digital campaigning compared with the Republicans’ paltry $4.7m. This approach allowed them to take the cream off the barrel, identifying precisely when and where to focus their efforts.

The 5 million people on Obama’s email list are just the start of what political strategists say is one of the most sophisticated voter databases ever built.

Not only did they have voters’ email address, they also had phone numbers, where they were registered to vote, a decent estimate of their household income and whether they’d opened a credit card recently. Obama’s camp knew how many children voters were likely to have and what they did for a living. And he knew what time of day people tended to get around to plowing through emails and respond to messages.

The ability to collect and analyse publically available data on a large scale allowed the Obama team to model behaviors before coordinating and targeting communications accordingly. With it they could, for example, predict which types of people could be persuaded by which forms of contact and content.

Although UK privacy laws limit what information organisations can collect and store, there is still a goldmine of open data at hand including any information we make freely available, such as social media shares, online comments, or YouTube posts. If you tweet your support for a specific party or policy, your comments will be publicly on display for anyone – individual, company or political party – to see.

Part of the reason this election looks so unpredictable is because traditional tools for forecasting are no longer up to the task. Most of the UK’s polls are conducted on a national rather than constituency level, failing to accurately reflect the election results. Our first-past-the-post system rewards concentrated support – think of it as 650 elections happening in May, not just one – yet our polling methods are configured disproportionally.

Big data should help solve this, with national polling results complemented by other available data that specifies the location of the respondent being polled. We can achieve a far clearer picture of the UK’s national psyche – constituency by constituency – through combining these diverse data sets in the right way.

Analytics can also make each pound work harder, with parties better able to focus on targeting the voters they need, helping limit overspending. Imagine that 10 seats will decide the election – by knowing where seats these are and what issues will sway the electorate, parties can save a fortune in terms of pound-per-vote spending.

But this is about more than 21st Century electioneering. Sophisticated data analytics can make a tangible difference at a policy level, creating a responsive government, more in tune with the electorate. Platforms such as Twitter and Facebook are already demonstrating the potential to crowd-source public opinion and channel the collective voice of communities. Big data can take this to the next level, forcing more responsive policies and representative politics.

Politicians who are in tune with their voters are far more likely to act on those issues most important to us. And as we make an increasing amount of data available to our politicians, we should expect them to employ the tools they need to use it wisely, and to our advantage.

One thing’s for sure, however – the party political broadcast is dead as we know it. All hail the hashtag.