This Startup Is Trying To Fix India’s Broken Election Polling

Last month, Arnab Goswami hosted a panel on Republic TV three days before the Assembly Elections in Rajasthan, called #Pollofpolls. Goswami invited two prominent Indian pollsters (those who conduct or analyse opinion polls), Pradip Bhandari of Jan ki Baat, an upstart in the polling game, and Yashwant Deshmukh of CVoter, one of the oldest pollsters in India. Bhandari and Deshmukh offered varied results—Bhandari gave BJP a slight edge, and Deshmukh gave Congress a sweep. Of the 199 seats up for grabs, Bhandari predicted BJP will win 83 to 103. The BJP ended up winning 73.

Even if we account for an accepted average margin of error, i.e. plus/minus 3%, Bhandari got his prediction woefully wrong, especially considering he had already used a wide net of 20 seats in his initial 83-103 prediction. This is more wrong than Hillary Clinton dancing at a wedding with Anil Ambani.

Exit polls, such as the one above, that are paraded as ‘opinion polls’ before elections, usually talk only about politics and people’s preference in leadership. Ideally, they should be used to monitor public sentiment on policy. “Democracy is not purely an electoral franchise. It’s hearing out what people have to say. Having these polls creates a voice for the people and a data bank for the voices, in terms of other topics as well not just politics,” says Sai Krishna Kumaraswamy, from the Consultative Group to Assist the Poor, a World Bank initiative. Opinion polls, if designed well, can be powerful tools for understanding public sentiment, which can lead to specific candidates being selected for office on issues of public interest, which can then change policy. Polls are important, but, as explained above, are mostly trash in India, and just too leadership-centric.

Delhi-based startup QNA (currently a website, with an app in the works) thinks they can fix this mess for the media and general Indian populace alike. “Think of us like Bloomberg for social attitudinal data,” says co-founder Kartik Desai. This is how QNA works: users sign up on the platform with an email ID and fill up three basic categories: age, gender and location. If they want, they can fill up three more categories: education, income/job profile, and identity (which includes religion).

After logging in, users are asked questions like, “Do you think Rahul Gandhi can challenge PM Modi after his recent elections sweep?” The questions link to news reports confirming the said election sweep, and gives users closed options to answer either ‘Yes’, ‘No’, or ‘Maybe’. They can answer questions, check analytics of others who have answered the same questions within the same age and gender, etc. And with each answered question, users get points, eventually (the feature is under development) racking up to a discount coupon on an e-commerce site.

Users’ privacy is protected by a tight encryption. In addition, QNA has no plans to sell their data. Ever. “Because the platform is for users, and needs their inputs. We have a self-interest in keeping the data credible and private to keep people returning to the platform,” Desai tells us. Those asking the questions, then, can narrow you down to, let’s say, 45-year-old Hindu male, residing in Mumbai, with income level of xxx.

Desai is the odd venture capitalist who wants to make money, and a difference. He has a degree in Finance from Wharton and a background in investment banking, but grew up in a politically-charged household—his mother is a sociologist and peace activist, and his father worked as a Senior Diplomat at the United Nations. He first engaged with politics while studying for his Masters in Public Policy at Columbia University, participating in the school’s first Indo-Pak delegation right after 9/11. He decided to merge his itch for finance and governance via an under-invested sector—civic tech.

“There is now citizen-to-citizen, citizen-to-government engagement, eGovernance, etc. QNA works as a platform to bring them together,” he tells me. “It can improve the quality of public debate.”

It’s a bit nerdy, but if enough users with diverse genders, location, income-levels, religion, and political views answer questions, we can get real-time data on what (mostly urban) India is thinking right before the elections, rather than wait for partisan political anchors, analysts, and news fucks to tell us the same. Surjit Bhalla, one of India’s biggest psephologists, gave all three states to the BJP in a resounding victory. He was wrong by about 30% in Madhya Pradesh alone.

Desai and his two co-founders intentionally didn’t include caste as a category because, although integral to the Indian electorate, it creates fractured data points. “It’s a subset of a particular religion,” says co-founder and content head Mirza Ali Khan. “We’ll have to include Hindu castes, Islamic sects, Catholics, Protestants, etc, which will reduce homogeneity.”

A key challenge is getting representation from all spectrums and locations, for which QNA wants to to partner with media organisations, think tanks and universities on the Left, Right and Centre. “There is no censorship of opinion on the platform. We’re only censoring abuses and un-parliamentary language,” says Khan. “And there’s no editorialising. Eg: When headlines break Modi’s demonetisation plan, QNA admin puts up one question, “This ‘thing’ has happened. Indians—What do you feel?” Someone will add in a question like, “Hey it’ll reduce black money, so it’s good. Do you agree?”

Cool.

So users get to answer questions on pressing matters, compare their answers with the rest of the populace, and earn coupons. What do the partners get?

“Our B2B will focus on civil society organisations and the media, as they have an intrinsic interest in publishing this information, versus a political party, who might want it on a proprietary basis,” Desai says. “If VICE wants info on culture, and runs a poll on LGBTQAI+ sentiment in Chennai just before the pride parade, within hours, there will be 150-200 people, hopefully more soon, answering relevant questions on our engine. We direct that question to people located in Chennai, and of the specified age. You have a sample size within 2-3 hours, whereas research would’ve taken much longer.”

Of course QNA charges a fee for the service. They don’t, however, plan to push the post in their users’ timeline, which currently toggles between ‘Recent’ (chronological) and ‘Trending’ (most engaging).

Discussing a poll on the feelings of third-generation farmers between the age of 18-25 in Rajasthan, or females in the same age bracket making below Rs 25,000 a month discussing employment policies, is far more insightful than say Kaun Banega CM. “The credibility of the media is at an all-time low and the technology-enabled echo chambers are polarised. If I already like Modi, I’m only likely to see positive news about him. Even leaving aside social impact, that’s a tremendous vacuum to fill for any venture capitalist,” adds Desai.

The vacuum is large, sure, but so is the challenge. QNA had a soft launch in October 2018, and currently has nearly 1,500 users. It’s not nearly enough, though. Desai understands this, especially after his work with WhatsUp Bharat, a 2011 startup meant to be the “Indian Habitat Centre for Inclusion”. “We’re trying to build a smart tank, which is efficient, scalable, disruptive and makes social impact,” he says. “We have small user targets—5,000-10,000—because we want them to be high quality samples. No one is saying it’ll be easy.”

Follow Parthshri Arora on Twitter.