When a project to empower left Congress stranded
Here's why Project Shakti, much-praised data analytics platform of Congress, failed to read the Modi wave.
The author of the email was Praveen Chakravarty, the Wharton-educated investment-banker and the head of the Congress data department since its inception in February 2018.
ET has a copy of the email and has reviewed it.
Chakravarty wrote to his colleagues, “It is a very tough day for us and perhaps even a dark day for the nation...I have spent most of the day today wondering what we could have done differently, why we were not able to catch such a big trend with all the analysis and the surveys and so on. I have concluded there is not much we could have done. I think each of (sic) should be truly proud of your contribution and effort. We fought for a cause and while it may not have been fulfilled yet, we shall continue the fight until it is fulfilled. We are not going to buckle down.”
That mail didn’t answer the question: why did Project Shakti, Congress’s data project, fail to read the Narendra Modi wave? Reading the mood was the data department’s core job. So, what happened? As one party functionary told ET, the department became an “echo chamber”, telling Congress president Rahul Gandhi’s office “what it wanted him to hear”.
There were other flaws, as unofficial post mortem inside the party shows. These include: the gathering of bad data, as previously reported by the news website Huffington Post, people working in silos, and giving feedback to Congress decision-makers, that Shakti insiders say, was “non-existent” or even “manufactured”, without secondary “validation”.
ET spoke to people who were part of Project Shakti as well as Congress functionaries. Most spoke on the condition they were not identified. Emailed queries to Chakravarty went unanswered at the time of going to press. Queries sent to the Randeep Surjewala, the head of the Congress' communication cell, and Divya Spandana, the party's social media head also went unanswered.
Shakti began last year when Congress, gearing up for national elections, started a data collection project that linked party workers’ voter IDs to their mobile numbers, through a simple short message service (SMS). The idea was to know who worked for the party where down to each individual’s address.
“Shakti, contrary to popular perception, is not an app. It was never an app. It is a platform,” said a New Delhi-based Congress functionary. The aim was to try and transform a loose volunteer-based party to a cadre-based organization. Via Shakti, Congress leadership could reach grassroots workers.
As Congress publicized it and much of the media reported, Shakti was a platform that allowed Rahul Gandhi to call a booth level worker, who would be energized by such open communication. Congress, therefore, would be energized.
One catch: for this to work, Congress needed good, clean data.
The data mess
According to Congress insiders, an early variation of Shakti was tried in Karnataka in late 2017, before May 2018 assembly elections in that state. Fake data immediately became a problem. “I think booth committees are harder to digitise because unlike bloc committees, they do not have a hierarchy (of officials). There was heavy pressure to collect data at the booth level, so 30%-40% of data had random names and random phone numbers,” said a Karnataka-based Congress functionary. Karnataka has around 55,000 booths, if Congress wanted 8-10 people per booth, 4-5 lakh members had to be found. As some insiders say, this effort was subverted by fake data.
No lesson was learnt, people familiar with Shakti told ET, and three months later the data analytics department was set up and the Shakti was born.
“How was this any different from the Karnataka exercise? Instead what leaders should have done is to put a mechanism for booth-level verification,” the Karnataka-based functionary quoted earlier said.
Instead, the party incentivized data collection, linking preferential treatment to data volume. ”District committee heads were catching every Tom, Dick and Harry, and were enrolling them,” another Karnataka functionary said. At one stage, the party even held a competition, in which the prize for most enrolments was a meeting with Rahul Gandhi, this functionary told ET.
Congress members from other states who spoke to ET agreed this was a fundamental problem. They said Shakti created a system waiting to be gamed.
All you had to do was buy SIM cards, take electoral data and send SMSes. With zero verification at the backend, this was a free for all, one party member said. The modus operandi, described to ET by two functionaries in different states, was this: “You register by sending your voter ID number to the party via your phone. Then, you go to someone else and request them to send a message with their voter ID number, and your phone number. This was like a referral scheme,” The New Delhi-based functionary quoted earlier in the story said, “enrolment figures became a means to ask for poll tickets.”
The data analytics department had also put up a giant TV screen at the Congress president’s house — 12 Tughlak Road, New Delhi. The screen showed a dashboard that assumed a new member was being registered every single time a mobile number was linked to a voter ID.
Coordinators of the data analytics department went to states and tell party members the Congress president had sent them, and the aim was to register new members. District presidents, one functionary said, had a target of 5,000 enrolments each, “they had no choice…they had to save their post.”
A classic example is from Telangana, according to Shakti insiders. One local Congress member hired an agency to enrol members for Shakti. A Congress leader in New Delhi said: “He essentially gave a contract to a call-centre while posing like an election commission official. He went to every single household and marketed himself like he’s come from EC. Basically phone liya, voter ID pooncha, aur SMS kiya (took their phones, asked for their voter IDs and sent SMSes). He claimed to have enrolled 12,000-15,000 members. The top leadership was very impressed with it”. This leader added: “It was all about self-aggrandizing. This was not about helping the party at all.”
The end result, party insiders say, was fake data of anywhere between 50%-70% across states. “Only 30-35% were your normal karyakartas (workers). These are hardcore, grassroots workers,” said another functionary. He made another point: Not all Congress workers are tech-savvy. Shakti was not reaching that kind of a worker”.
The one semi-exception was Maharashtra, Congress data people said. Shakti data collection was somewhat better in terms of effectiveness, even though there were slippages. “We filtered this data well, scrubbed it and identified bad data. We didn’t focus on the numbers game, just on the quality game,” said Abhijit Sakpal, general secretary of the Maharashtra Congress, and state-coordinator for Shakti. Congress plans to double down on Shakti usage during the assembly elections in Maharashtra October.
Problem of silos
Before shortlisting advertising agencies for general elections campaign, Congress wanted to test out certain videos its publicity committee had commissioned. These videos were sent to Shakti members via SMSes, with YouTube links embedded.
But these videos got around 20,000-25,000 views, far from what’s considered “going viral”. “The first warning signs came when Divya’s [Spandana] team (the social media department) refused to put out those videos on any platform…because they hadn’t gone viral,” said a senior Congress functionary.
He added: “This was because the data analytics department and specifically Praveen [Chakravarty], its chairman, was never forthcoming with the data. There were suggestions of an oversight committee for Shakti…that idea was scuttled by Chakravarty. [Kopulla] Raju was very keen on more oversight. So was Jairam Ramesh.” Raju, a former bureaucrat, works in the Congress president’s office as the head of his core team.
Concerns over the centralisation of the data department’s operations were flagged on multiple occasions to the Congress president’s office, the New Delhi-based functionary quoted earlier said, but to no avail. “Few had any inkling about the data.”
“Access to Shakti was tightly controlled, and even within his team, one member didn’t know what the other was doing,” said the functionary quoted above. Such was Chakravarty’s influence inside the party that a consultant working for a senior Congress leader described him as the “Congress’ super-president for six months.”
The data analytics department also sent an email written by William Chitla, a close associate of Chakravarty and a Bengaluru-based IT entrepreneur, which described “Shakti Data Security Policy”. ET has a copy of the email and has reviewed it.
Chitla wrote, “As per the new “Shakti Data Security Policy” we will not be providing AC (assembly constituency) level Shakti registered users data for download. Instead, we will be providing booth-level Shakti registered users data in incvidya.in and incshakti.org to download. The New Shakti policy is as per CP’s instructions.”
This raised some alarm bells within the party. The senior functionary quoted earlier said, “A private person who is not even an office-bearer of the party is sending directives to office-bearers of the department. This person, along with Chakravarty, hijacked everything about the party. Without having the authority of any kind.”
When ET contacted Chitla over the phone, he said, “I don’t have any idea [about Shakti]” before cutting the call. While is unclear if Chitla is a member of Congress, he did have an email address linked to the INC data analytics department. All of this, multiple people said, was down to Chakravarty’s direct access to the Congress president.
Party insiders said the social media department didn’t get data they requested from Shakti, and the former therefore ran a parallel data analytics operation. The department even commissioned a survey by British polling agency YouGov. “They (Congress social media members) kept asking for stuff. But the answer was ‘no’ all the time. There was no spirit of cooperation at all” , a Congress mid-tier leaders said.
The consultant quoted earlier added that senior leaders did not speak much about Rafale – because they were at loggerheads with Chakravarthy. Rafale, this person added, was fed into Rahul Gandhi’s thinking by Chakravarty. “It was almost as if there was a cosy club at Gurudwara Rakabganj, which met and prepared a dossier on the mood of the people and gave it to the Congress president. A lot of senior leaders, who have a better ear-to-ground sense, refused to campaign using Rafale as part of their speeches,” the consultant said.
Shakti was in a silo of its own, Congress members said. But perhaps the bigger problem was its influence in political decision-making.
The echo chamber
Shakti was originally conceived as an internal communication platform. But with data collected, it became a tool for on-ground surveys. Congress leaders told ET it was heavily used for feedback. There were two problems with this. One, bad data. Second, many genuine party workers didn’t give genuine feedback. It became a giant echo chamber, one Congress member said.
“No committed karyakarta of the party will tell you the reality,” this member said, before adding, “It had limited use.” The greatest mistake, he felt, was using Shakti to identify the narrative. “How you end up using the platform is very important. It was a great idea…we desperately needed an internal communication platform. But to gauge the mood of the nation through Shakti was inherently flawed.”
But what was equally flawed, multiple functionaries felt, was that political decisions were being taken based on these surveys. Some surveys were also conducted by external agencies, Congress insiders said, but alleged that details of who conducted these surveys and how were “kept a secret by Shakti people”. Another senior party functionary said: “On what basis did we conclude Maharashtra is an angry state? Or that Congress will triple its tally in the general elections? Political decisions were taken based on these surveys.”
Speaking of Chakravarty, he said: “This was, after all, a man who called 2014 a 'Black Swan' moment that will never be repeated. How could he not read this sentiment?"
Some Congress members allege even decisions on whether or not to get into an alliance was influenced by Shakti surveys. ET could not find more verifications for this claim.
Another functionary pointed said, of Chakravarty: “I think he did a good job from a focused perspective when it came to data collection. So what if there was wrong data? It (data collection) is always a messy affair. Problems started when he allegedly relied on it to do surveys. He doesn’t have a market research background, and the part he didn’t know, he shouldn’t have done it himself.”
He gave the example of Gujarat. He said the model used by Chakravarty and his team indicated Congress may win upwards of 15 seats in the state. BJP swept Gujarat, winning all 26 seats. “What was the basis for this prediction? What was the model? Chakravarthy never shared how he came up with these numbers,” this functionary said.
Another senior functionary said “red flags were raised over bad data being sent to the party president’s office…but no one paid heed to it”. “Hardly any of us were involved in any decision-making process…how could we have given inputs or checked the data.”
The idea of Shakti remains valuable, many functionaries said, but the use of it was deeply flawed. They say the critical need was for oversight. The consultant quoted earlier said was severe, calling Shakti a “colossal failure.” “It was as simple as this: an assumption was made that there was an anti-BJP wave and that voters will automatically vote for Congress.”
Chakravarty’s May 23 email ended thus. “For the way forward, once things settle down, I will have a conversation with the Congress President to draw up a vision and plan for the department. Until then, it is only appropriate that we put things such as Shakti registrations on hold.”
ET had sent three emails to Chakravarty during the course of reporting, one to a wrong email id, which promptly bounced back and landed in this reporter’s junk email. The first email, on the 12th of June, contained 13 questions (including a numbering and a grammatical error) on everything this reporter had researched, including conversations with multiple functionaries of the party, at various levels of its leadership. These questions ranged from what went wrong with Shakti, and the data analytics department, to William Chitla, to what our sources told us about the department working in silos, and the feedback mechanism that Shakti acted as for the party, among others. A follow-up email was sent as a reply to the questions on the 14th of June, saying that we were awaiting his responses to the questions sent earlier. When the story went to print on Sunday night, ET did not get a reply from Chakravarty, or two other Congress functionaries – Randeep Surjewala and Divya Spandana, whom ET reached out to on the 13th of June.
However, following the publication of this story on 17th of June, Chakravarty, on his Twitter profile, put out a signed statement. He wrote, “A recent gossip column about the activities of the Data Analytics department of the Congress party during the recently concluded Lok Sabha elections has been brought to my attention. Every line in that column is a lie. Equally, other similar stories based entirely on unnamed sources masquerading as investigative pieces, are plain rubbish. These are obviously, mischievous, absurd, and sadly, an utter disgrace to Indian journalism. My department and me continue to function actively from our offices at the All India Congress Committee headquarters, as usual.”