India is one of the largest democracies in the world, and Indian elections are a sight worthy of attention. The grand and vibrant elections, however, are not easy to win. Data analytics and AI are major factors in the back end of these elections. 

With piqued interest, Analytics India Magazine got in touch with VoteBridge‘s founder and CEO, Vasudha Singh, who had previously served as a senior consultant for the governments of West Bengal and Madhya Pradesh. She also served as a state lead for the AAP in the 2020 Delhi State Election and the Shiv Sena in the 2019 Maharashtra Legislative Assembly Election. In addition, she has experience working in several elections and has over six years of expertise managing campaigns and providing strategic research consulting in both the governmental and political spheres.

AIM: What are the different data points (of voters) collected for analysis? Do data points like caste play a major role? 

Vasudha: Typically, the analysis uses publicly accessible data from election commission websites, such as gender breakdowns, the number of voters who have registered, etc., for future elections. One might examine the Polling Percentage, Booth Level Votes Polled, NOTA, etc., to analyse data from previous elections.

Coming to the intersection of caste and data, winnability is the key area to investigate. It might be challenging to understand how caste influences voting behaviour. To win an election, one would typically require between 40,000 and 60,000 votes. In no constituency in India would there be 60,000 voters exclusively from certain caste demography, and even if there were, not everyone would vote for the candidate from the dominant caste.

Another scenario: If a caste dominates a seat and all parties choose their candidates based on that caste, the caste votes would be split. In order to respond to the issue, we may say that, in our political experience, the most important aspect is not caste but rather a candidate’s potential to win—which depends on a variety of criteria in addition to caste, such as their performance, popularity, accessibility, etc.

When talking specifically about voter data and subjective inputs from voters, any interested parties will have to do an on-ground/calling survey and acquire information through direct consent.   

AIM: How long does it take for a data team to formulate a plan? 

Vasudha: The size and significance of the project that a data team has taken on will determine the plan. If a team wishes to choose the best itinerary for a leader’s State-Wide Yatra, they may need to take into account data from the state’s last six elections, the strength of the cadre on the ground, major issue inputs, best travel routes considering time and resource restrictions, etc. This will be a lengthy procedure that might go on for a month or longer. 

However, a booth-level study of one polling place or station for a single candidate might be completed in a matter of hours.

AIM: What has changed, or evolved, in political campaigns since the US presidential elections of 2008 in terms of data analytics? 

Vasudha: Since the US elections in 2008 and generally all elections in India post-2014, data analytics has been used by most parties and top leaders. More and more decision-making in the political system is now moving towards facts-based understanding as opposed to anecdotal insights by a group of close aides of leaders. 

Leaders now understand the importance of collecting data from their party workers, introducing an internal feedback system, and 360-degree reviews from their own cadre. Rallies, campaigns and ticket distribution, in many cases, are decided through analysing several data points such as popularity, key voter groups, the strength of cadre, perception of leader, crowd-pulling power and more.

Political campaigns have also become all about easily quantifiable numbers which are released by parties and the political consultancies that helped them. For example, Stalinin 7 Urudhi Mozhigal reached more than 75 lakh families, Arvind Kejriwal’s Guarantees promised an INR 16k+ crore Budget for Education, Shiv Sena distributed INR 4.27+ crore Shiv Bhojan Thalis, Nitish ke 7 Nishchay promised 100% electrification of villages, INC in UP 2022 promised and delivered tickets to 40% female candidates. 

This shift in data occurred after 2014 when political parties in India began using large numbers to emphasise their accomplishments, pledges, or endeavours. Political parties may now measure their efforts and share them with the public, which would otherwise be lost in the ‘festivities’ of the Great Indian Elections—thanks to the development of data and technology.

AIM: How effective is data analytics in states like the UP? 

Vasudha: Data analytics cannot be political state-specific, in my opinion. Any analytical model one creates can only fully achieve its goals if it is accurate. The number of important actors in the state, rather than the state itself, is what determines accuracy.

It is simpler to anticipate voting patterns in a state with a big population and a dual-party dynamic, such as UP because the margins of win/loss are significant and can be determined using a mix of previous election data and current poll estimates. While in places like Goa and Tripura, where there are four political parties involved, and the win/loss margin might range from a few hundred to thousands of votes, the models used to forecast the outcome would have a lower accuracy rate.

AIM: Which party is the most “tech-savvy” in terms of the use of data analytics in their campaigns? 

Vasudha: In practicality—AAP, BJP, INC and regional parties as well are all tech-savvy. They are all aiming to increase their ability to make judgments that are supported by evidence, which will enable them to use their resources as efficiently as possible for maximum vote conversion. Nevertheless, the depth and accuracy of the models used by different parties vary to differing degrees.

The party with the best chances of winning this game is the one that is most receptive to the idea of fusing technology and politics.

AIM: How do you think the collected data will be used after the elections?

Vasudha: Data collecting for specific people or votes is often not done at any centralised level. A prediction model is created using publicly available data from previous elections. In regards to the voter data gathered during a survey, the conclusions drawn from a field investigation lose their validity after three months, and they do so much more swiftly as elections draw closer. 

Even in that scenario, the goal of all data collection is to identify the winner. Therefore, pre-election data of any kind is scarcely relevant after the election. Every election sees revisions to even the highly sought-after voter and booth list. Yes, however, parties do gather and routinely update the information about their party members and maintain contact with them during the election.

AIM: What are the different sources of data collection? Does social media play a huge role? 

Vasudha: In India, they may be found on the website of the Election Commission, in-person or over-the-phone surveys conducted by organisations, and on social media platforms.

Social media is undoubtedly one way to get information, but it also presents a significant obstacle. It is very challenging to determine whether the opinion and input of a given account are valid or useful due to the anonymity of the account, doubts about whether the account is owned by a real person or a bot, uncertainty regarding whether the input is from a constituent or even a resident of the state, and other factors. 

Additionally, because there is a poor and diverse digital adoption among voters in India, political consultants rely more on conventional techniques of gathering information than they do on social media.

AIM: Has Prashant Kishor altered the field of managing political campaigns? 

Vasudha: Indira Gandhi, Rajiv Gandhi, Atal Bihari Vajpayee, and other leaders in Indian history have a long history of professionalisation and the usage of experts. However, earlier firms and people used to function more as a “close aide” to the party at its centre than as an extensive implementation unit.

But what Prashant Kishor and the I-PAC Model brought to the nation was a whole industry where anyone could come and take part in the country’s electoral processes without having any political ties and/or the desire to contest elections. I-PAC has contributed to the political ecology with its sizable network of field agents, ability to comprehend local problems, and goals for enhancing conversion at the village and booth levels. Even if Bitcoin was not the first cryptocurrency, it has been a successful proof of concept for the use of data in politics.