Big Data in Political Campaigns around the World

Authors: Shikhat Karkee, Sabin Bhandari, Georgi Panev, and Dennis Luepkes

Around the world Big Data is used to predict Political Elections, as well as to make sure that the allocation of advertisements is used to the best of their ability.We have collected examples from all over the world that have used Big Data in their political campaigns.

The first Prime Minister to use Big Data:
15281038_1378419628865551_374506396_nNarendra Modi’s use of big data analytics where Congress lost so heavily to the Bharatiya Janata Party (BJP), making him the prime minister of India.

For example, the size of the Indian electorate. With 814 million voters, the sheer volume of data of India’s voting population was perhaps the largest obstacle. The second was the variety of data – India’s voter rolls in 12 different languages and 900,000 PDF’s amounting to 25 million pages made for a heterogeneous, non-uniform and deeply diverse information set. Finally, the veracity of the information was often questionable – one report noted that some voters were listed as 19,545 years old, and others a confounding 0 years old. Name overlapping (there are 327,000 women named “Sita” in Bihar alone) only further complicated the process.

Milind Chitgupakar, the Chief Analytics Officer of Modak Analytics (a company that has built India’s first Electoral Data Repository), and his group of 10 data scientists used everything from heat maps and data visualizations to complicated machine learning algorithms to sift through the volume, variety, and veracity of the data before making it available to political parties in India.

Despite these challenges, the rewards – as Modi has clearly demonstrated while employing this data to “drive donations, enroll volunteers, and improve the effectiveness of everything from door knock to social media” – are significant. BJP’s website, for example, planted cookies on all computers that visited its site, and then used information about these users’ further internet activity – i.e., the sites they visited after BJP’s – for customized advertisements:

“If you move out of the BJP website and visit a website for bikes followed by a search on jobs, the algorithm will make the inference that you are a young male from a constituency, say Delhi, who is currently on a job hunt. What happens next is when you visit a job searching portal like, this system pops up a contextual ad for you like ‘jobs in Delhi’. The BJP banner which is just below the results will tell you ‘There are no Jobs in Delhi. India deserves better’.”

Use of big data in Australian Political Campaign:

picture1Malcolm Turnbull, who has 282,000 ‘likes’, is a good example because we can see a personal message and him talking about plans for jobs and growth. While Bill Shorten’s page has fewer than half the ‘likes’ of Malcolm Turnbull — 118,000. But he’s using the medium here to put out negative messages about Turnbull and this is quite a clever way of doing it because it’s cheap, it’s quick and it bypasses traditional media.

But with a bit of luck for the political party it will also be picked up by mainstream media, and this is important because it saves them big dollars and it reaches maximum viewers by traditional and social media platforms. 

How Big Data played a role in the 2012 election:

safe_image-phpTo understand the elections of 2012 and why Obama managed to get elected again, we must first mention a few important facts about the elections that happened in 2008. A guy called Dan Wagner who is famous for his system of making predictions about outcomes of elections based on threating people as equal and not dividing them by groups, categories or any other factors, joined Obama’s campaign team. He used his model to score people based on their likelihood to vote for Obama. The scores were determined by mainly phone interviews as between 5 000 and 10,000 “so-called short-form interviews” and around 1,000 “ long-form” ones resembling a traditional poll were executed every week (Issenberg, 2012). Complicated algorithms operating with around 1,000 variables were applied to understand individual-level predictions based on people’s opinion, their registration records, consumer data warehouses and past campaign contacts (Issenberg, 2012).

Back to the elections of 2012, when the campaign new started Obama’s team knew the names of all 69,456, 897 voters who elected him in 2008 and they planned to reach them personally in order to have their votes again. They also took data from commercial warehouses concerning all adults eligible to vote and intended to contact them as well via email. The content of emails was developed by first making empirical research on the best themes and language to use. For instance, one such topic to be investigated how to be discussed with the public is health-care policies (Issenberg, 2012).

Another crucial step they took was to have personal interaction with people by using volunteers to talk with them and persuade them vote for Obama. More than 500,000 Americans were reached in that way (Issenberg, 2012). It turned out to be a very successful approach since “Millions of Americans heard from other Americans about issues that mattered to them. Those conversations were more powerful than the billions of dollars spent on TV ads” (Trippi, 2013).  Other forms of advertisement were used as well. For example, huge amount of money was spent on TV ads. TV channels watched by people who were not fully decided who to vote for had to be reached to convince those people to choose Obama. Since TV providers don’t just allow everybody to see the histories of their users, the Obama campaign asked for a repackaged version of the data in such a way that it does not violate the firms’ privacy standards (Issenberg, 2012).

Traditional polls were also used as a mean of collecting data and using it to change the image of the campaign and put more focus on some topics. Targeting specific groups was possible because of the polls as well as the results collected by the other means of data collection (Issenberg, 2012). More interaction between voters and politicians was established by the usage of social media (Gerodimos and Justinussen, 2015). It surely became a significant part of people’s daily lives as they searched for more and more information regarding politics on social media web sites. The exact messages were carefully designed and shaped.

How Big Data played a role in the 2016 election:

donald_trump_14235998650_croppedIn the election, this year there was a lot of turmoil even 2 years out from election day due to the use of social media. Big Data played a big role in this election from being misused by Hillary Clinton and being drastically overused by President-Elect Donald Trump. For instance, “Clinton’s campaign used a custom algorithm called Ada that staff fed “a raft of polling numbers, public and private” to help Clinton’s team decide where they should dedicate their resources” (Vanian, 2016). This severely hurt the Clinton’s campaign as they did not focus in the key states that the data said she would guaranteed win. “In Pennsylvania, which polls projected to win, … Trump ended winning the Keystone State by a thin margin” (Vanian, 2016). Meanwhile Trump “… seemed to base their decisions from the emotions of crowds attending Trump’s rollicking campaign events” (Vanian, 2016). This means that while Clinton focused her advertisements, Trump attacked a wide variety of places. Trump states, “I’m not saying I love it, but it does get the word out. When you give me a bad story or when you give me an inaccurate story…” (Trump as stated in McCormick, 2016). However, Trump’s late night tweeting campaigns are not the only thing giving him a lot of attention, but one cannot forget about the false news on social media sites. Per Parkinson, “The influence of verifiably false content on Facebook cannot be regarded as “small” when it garners millions of shares. And yes, it runs deep. The less truthful a piece is, the more it is shared. In Zuckerberg’s follow-up statemant, he seems to have shot himself in the foot, by saying it was “extremely unlikely” fake news on Facebook had an impact on the election, but also boasting that Facebook was responsible for 2 million people registering to vote” (Parkinson, 2016). Thus, Big Data had a huge impact on the 2016 election whether it is the data analytics that predicted Hillary Clinton to win, or the social media tirades and false news that led to President-Elect Donald Trump winning.

Other Examples:

Norway is consolidating its voter registration records, and these may be available to nongovernmental entities. In Britain, Canada, Australia, and several other democracies, the parties or candidates have access to electronic voter registration records to allow them to more effectively campaign in elections. The parties can then add other data from party records to develop a richer profile of individual citizens. Some reports of the October 2015 Canadian parliamentary elections thus claimed it was Canada’s first Big Data campaign (Ormiston 2015).


Political campaigns amass enormous databases on individual citizens. This data-driven campaigning gives candidates and their advisors powerful tools for plotting electoral strategy.

Based on data acquired from citizens, the campaigns can predict with greater accuracy which citizens will support their candidates. Information regarding citizens who donate, volunteer, and subscribe to email lists is available to campaigns. Sophisticated campaigns develop and use voter databases that contain a range of detailed information on individual citizens which increases cost effectiveness of communicating with citizens, a broad range of organizations do and will employ the technologies.

The supply of quantitatively oriented political operatives and campaign data analysts has increased as predictive analytics has gained footholds in other sectors of the economy like banking, consulting, marketing, and e-commerce. To reduce the need for individual campaigns to spend scarce funds purchasing citizen information from commercial vendors, the national parties have decided to construct, maintain, and regularly augment their own voter databases (McAuliffe and Ketten 2008, p. 280-287).


The public discourse on campaign data has been largely speculative and somewhat hypothetical which raises concerns about the personal privacy of voters (Duhigg 2012). The disinformation caused by targeted news on social media is proving fatal to truth in political discourse. In election, if polls don’t reach all likely voters, there could be data insufficiency which results in vague data.

In addition, data leakage mainly due to hacking, deriving conclusions from erroneous data patterns and too much reliance on data are major threats on big data used in political campaigns.


With all this data flying around, it’s hard to know how to digest it! Because political campaigns run like mid-sized companies, they often have data that is so complex and distributed that it needs a dedicated team of specialists to clean and maintain it. Unfortunately, this can create a bottleneck between the IT operation pulling the data and the overall campaign team, which needs quick access to data while hot on the campaign trail.


Gerodimos, R. and Justinussen, J. (2015). Obama’s 2012 Facebook Campaign: Political Communication in the Age of the Like Button. Journal of Information Technology & Politics. 12 (2), 113-132. Retrieved on 28.11.2016 from

How Campaigns and Companies Use Data to Win the Race. (2016). Retrieved November 28, 2016, from

Issenberg, S. (2012). Obama’s Data Techniques Will Rule Future Elections. Retrieved 27.11.2016 from

Issenberg, S. (2012). How Obama Wrangled Data to Win His Second Term. Retrieved 27.11.2016 from

Issenberg, S. (2012). How Obama Used Big Data to Rally Voters. Retrieved 27.11.2016 from

McCormick, R. (2016). Donald Trump says Facebook and Twitter ‘helped him win’ Retrieved November 28, 2016, from

Parkinson, H. J. (2016). Click and elect: How fake news helped Donald Trump win a real election | Hannah Jane Parkinson. Retrieved November 28, 2016, from

Trippi, J. (2013). TECHNOLOGY HAS GIVEN POLITICS BACK ITS SOUL. MIT Technology Review. 166(1), 34-36. Retrieved on 28.11.2016 from

Vanian, J. (2016). How Bad Polling Data Fooled Everyone Except Donald Trump. Retrieved November 28, 2016, from

Why Big Money and Big Data Win Elections. (n.d.). Retrieved November 28, 2016, from

Why Winning Politics Is Now Tied to Big Data Analytics. (2016). Retrieved November 28, 2016, from


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s