The age of Flipkart


A decade ago, Sachin Bansal and Binny Bansal were two young twenty-somethings with a plan—to set up an online bookstore. Neither they nor the rest of the country could have guessed that this store, started with an investment of just Rs 4 lakh (just over $6,000), would go on to become the flag-bearer of the Indian e-commerce industry. Flipkart did just that.

Today, the company is valued at over $11 billion, hosts about 1 lakh sellers, has a registered user base of around 100 million people, boasts a catalogue of around 80 million products, accounts for over half of India’s online smartphone sales, claims to have captured three-fourths of the online fashion market, and is locked in a fight-to-the-finish with global behemoth, Amazon.

For much of 2016, Flipkart was on the ropes as Amazon out-spent, out-marketed and out-innovated its much younger counterpart. Last year, Amazon pipped Flipkart to become India’s number one online retailer. Never one to back down, Flipkart came back swinging and by the end of Diwali, the country’s biggest spending season, it had regained its leadership position, even if by a hair’s breadth. As it completes a decade in the business, YourStory takes an in-depth look at what Flipkart is doing to remain India’s biggest e-commerce player.

Innovation Story

Four Pillars

The Interview

Flipkart DNA

Flipkart looks deep and re-discovers its innovative soul

Hip and trendy mobile phone cases and covers are not big contributors to gross merchandise value (GMV), so why does Flipkart’s Vice President of Engineering call it a very important category? True, they’re fast-moving products and customers keep coming back for more. But that’s not it. For hundreds of thousands of first-time online shoppers from beyond the metros and big cities of India, a mobile phone case priced around Rs 300 is their first transaction. The low price and minimal chance of in-transit-damage makes the product a relatively risk-free first purchase, says Amar Nagaram, Vice President of Engineering at Flipkart.

Each of these transactions results in priceless data for Flipkart. So what does Flipkart do with this data horde? There is a series of actions that a consumer carries out on the portal, which lead to a transaction. Website analysts call this the Golden Path. Flipkart’s top-of-the pyramid users—those with plastic money—typically follow this path. But what the company’s product and engineering teams wanted, was to discover the customers in the low average selling price (ASP) categories like cases and covers, who also follow this path. The challenge was that these were customers who use cash; their digital footprint is close to zero, and their transactions on Flipkart constitute almost their entire digital trail.

So Flipkart’s engineering team built an algorithm and identified customers from this population who do follow the Golden Path—the ones who always accepted their cash-on-delivery orders and never cancelled or returned products. To begin with, Flipkart gave these buyers small amounts of credit to test the hypothesis that though they might not have a credit score or a credit history, they would nevertheless repay on time. What was the reason for this almost-Big-Brother-ish exercise last year? For the answer, we need to go back in time a bit.

The trip to Darjeeling

It was the summer of 2016 and Flipkart was conducting yet another sale, this time on a few large-appliance categories. While the discounts were attractive as always, the company found that several offline stores that weren’t offering any discounts were recording better sales. The reason, they discovered, was the no-cost EMI they offered, which made the purchase of an expensive air-conditioner or a refrigerator a bit easier on the wallet for buyers. Flipkart’s product and engineering team knew that they could replicate the no-cost EMI model online but they wanted to do one better.

The affable Amar gets animated only when he’s talking about two things: one is iconic techie and entrepreneur Steve Jobs, and the other is how Flipkart goes about solving its engineering and business challenges.For customers who have plastic money, salary slips, bank statements and other such proof of income, it was just a matter of integrating with banks to offer this. The company also tied up with financial institutions like Bajaj Finserv and ensured all the paperwork could be done online in a few simple steps. But what Flipkart really wanted to offer was the no-cost EMI option to the population that operates with cash. That’s where their engineering team got cracking.

And animated he was when he spoke to us about how the team went about finding a solution to the challenge of offering no-cost loans to cash-shoppers.

At Flipkart, a lot of things happen in parallel. The day the idea struck us that (the idea of) no-cost EMIs needs to be brought online, the engineering team started working on it. At the same time, another team went on the road to figure out what else we could do,” recalls Amar, who has been with Flipkart since 2012.

The team on the road travelled to around 10 towns in states like Rajasthan, Madhya Pradesh, Uttar Pradesh and Assam. The tourist town of Darjeeling in West Bengal was one of the towns on their itinerary. “We did a complete user study there to understand how Darjeeling shops,” says Amar. What they discovered was that most of the cash economy population in the tourist town earned their money during the peak summer tourist season, and in the fortnight before Diwali, buy whatever the family needs. Every two or three years, when the family needs to splurge on a big purchase like a refrigerator, they skip buying new clothes for Diwali.

The team found that this population, while not having much of a digital presence, had made some small transactions on Flipkart. If they had a no-cost EMI option, they could well make the transition from mobile cases and covers to larger purchases. And that’s what brings us back to the small bits of credit that Flipkart experimented with.

Chart 1: Flipkart's funding and valuation over the years

“Initially, customers were sceptical as they were the kind of people who did not opt for loans. But once they got the essence of it—that they do not have to wait for three years to buy a television set—we saw conversions happen,” recalls Amar.

Today, Flipkart offers buyers the no-cost EMI across a slew of products ranging from large appliances and phones to even designer fashion. And every time a customer buys something in one these categories, the algorithms assess if she is eligible for such a loan and offers it at the time of checkout.

The business of innovations

Innovations like this are important for Flipkart. When it started out, the major challenge was to convince people who were already transacting online, mainly for airline tickets, to also shop for other things online. Cash-on-delivery (CoD), which Flipkart introduced in 2010, was the game changer as it convinced users to try out the site at least once. Today the challenges are very different. While around 40 million Indians actively shop online, the industry that was more than doubling every year has plateaued—the Indian e-tail industry grew at a relatively negligible 12 percent in 2016 to reach $15 billion, according to advisory and research firm RedSeer Consulting. In 2015, the growth rate was a whopping 180 percent.

Flipkart too needs to grow its sales dramatically. Its monthly gross merchandise value (GMV) has been hovering between $300 million and $400 million in non-sales months, according to an industry insider who consults with both Flipkart and Amazon. During large sales events like the October Big Billion Days, GMV doubles.

Flipkart is also constantly facing the heat from a high-spending Amazon, which is taking a hit of around $1 billion each year on its operations in India but showing no signs of slowing down. Flipkart and Amazon, not to mention all other e-commerce players, are competing for a share of wallet of the same 40 million online shoppers.

“Flipkart figured out a while ago that whoever wins that battle (of discounts) will end up losing the most money,” the person told YourStory on condition of anonymity. And worse, these discounts were not leading to loyal consumers. Several people YourStory spoke to said Flipkart’s monthly spends stood at around $50 million last year. Its goal is to slash that number by half.

A townhall in progress at the Flipkart office

A feature like the no-cost EMI is part of its strategy to lower this burn rate—the umbrella term for spends on marketing, discounting and other measures a company takes to get a customer to shop with it. Flipkart is focusing on solutions that will net it loyal customers who are not interested in just discounts.

You need to keep innovating to increase the market size you can cater to. The first set of customers may be coming on (the basis of) price, selection and product. But the competition can replicate this. Which means that fundamentally, you have to offer something different to customers, which ensures customers see a value in coming to your platform, and that value has to be something other than price. That’s why we are focusing on affordability in a big way,” says Ayyappan R, Senior Director heading the Mobiles and IoT category at Flipkart.

The company is also looking beyond the base of 40 million active shoppers. A 2016 study by Google and research firm AT Kearney estimated that India will have a total of 175 million online shoppers in 2020. However, the needs of the next set of online customers could well be very different from those who live in large cities, have access to a wide selection offline and own credit cards. “The challenge we have right now is that the new customer doesn't even have an email address, has never seen a website, and has not seen any form of the internet other than WhatsApp. How are you going to make this guy to get through 37 million products that we have on our catalogue and show him the most relevant product that he is looking for, even though he didn't tell us his exact intent?” explains Amar.

The third challenge is one of perception, according to the company. “Pretty much every major innovation this industry has seen in the last 10 years has been pioneered by Flipkart, (and it) has not been given credit for that. It is a perception issue that the foreign giant, the number two player, is doing it (innovating),” says a Flipkart spokesperson.

However, the perception stems from reality. It was Flipkart’s early business and tech innovations like cash-on-delivery, 30-day no-questions-asked returns and the exclusive mobile phone launches (see Chart 2 ) that helped it capture customers’ imagination. However, in 2015 and 2016, Flipkart’s audacious moves like an app-first strategy failed and management churn increased. It spent most of the second half of 2015 and all of 2016 recovering from those fiascos. In the meantime, Amazon surged ahead with experiments like its FMCG and grocery play Amazon Now, and scaling up Kirana Now, through which Amazon uses local mom-and-pop stores as pick-up and drop-off points.

Time to play to its strengths

We are playing to our tech strengths because that is something we are very good at. Whenever we do something manually because we want to pilot it quickly, we do it badly. We get better results when we go the tech way,” says Ram Papatla, Vice President, Product Management at Flipkart.

At any given point in time, Flipkart has about 100 experiments running. Some might be the result of its data team having identified patterns, or the product team having spent time with customers. Others might be the result of the business team facing specific challenges and needing solutions, or an initiative that the tech team is leading. Ram oversees them all. He believes that constant experimentation will be the name of the game in Indian e-commerce for the next few years.

While a customer’s journey from discovery to payment online may seem a fairly straightforward process, there are many potential points of friction—from discovering the exact product a customer wants despite not knowing its correct name, spelling or brand, to adding credit/debit card details each time a payment has to be made if a customer does not want to save these details online.

“The fundamental thing about e-commerce in India is that no one knows what the playbook is. A series of experiments are needed for the next few years. From how customers come on to the internet (using their smartphones) to being able to understand what to look for and once they understand that, to be able to trust that they should buy it (online)—a lot of things need to happen over the next few years for this to become completely friction-free,” says Ram, who joined Flipkart in 2015 after working with Amazon as its Director of Product Management and Marketing for a year.

Most of Flipkart’s innovations are focused on these areas—how easily can a customer access the site on the web or via a phone? How easily can they browse through the site and find products that they want to buy? How can their decision to buy become an easy one, and finally, how simple can the process of paying for a product become?

“What we have been focused on for the last several years is to make sure that decision-making is more important compared to business performance. So, let's make sure that everything is focused on consumer decision-making. Once you do that, business will perform,” says Ram.

Chart 2: Flipkart has been responsible for many industry-first innovations that directly led to the growth of Indian e-commerce; Graphics by Aditya Ranade

Helping the customer buy—with AI

The progressive web app Flipkart Lite, launched in late-2015, is a website that has all the features of an app. This makes accessing and fully experiencing the portal easy even for customers who do not have the Flipkart app and are browsing on a 2G network. Once the customer is on the site, the homepage is personalised based on multiple parameters like past behaviour on the site, device patterns and what’s trending.

Many users wanted the app to be faster, and the easiest way to do this was to reduce image size. Sure, Microsoft and Google have already executed this, but they didn’t have to deal with the problem that Flipkart faced. When an image is compressed, so is the model’s face and the result can put off customers. So Flipkart’s team went about building a computer vision solution where it compresses the rest of the image but not the model’s face. In some cases, the face is rebuilt (Flipkart’s tool can recognise 80 skin tones). The tool can detect duplicates, match similar products, extract attributes of even Indian ethnic wear, clean up images, identify not-safe-for-work images and even understand sentiment in text, like in a review.

The moment of truth is when a customer moves from browsing to making a purchase. A bunch of tech innovations are clustered around this. For instance, the Flipkart team has linked children’s shoe sizes to age, leading to improved sales conversions. The company declined to share the exact impact of such tech solutions in terms of actual sales numbers.

Where offline has a distinct advantage over online is a salesperson who offers real-time responses to a customer’s queries. Project Mira, Flipkart’s Artificial Intelligence project, is an attempt to converse with its users. While it is text-based for now, the intention is to integrate it with voice and make it work in regional languages too. This is expected to help customers with their purchase decision.For now, the company has unveiled Brand Chat for some products, where customers can ask questions like, “Do I need to use oil to make French fries in the air fryer?”

For now, this is serviced by an on-demand team, but Project Mira’s super-bot is expected to eventually automate responses.There’s one more feature that came about as an extension of the no-cost EMI option. Amar recalls how the team saw lower-than-expected conversion, especially in its stronghold of mobile phones, and decided to get to the bottom of the mystery—why were customers not buying despite the attractive discounts and exchange offers? It turned out that for many customers, the exchange of old phones was the obstacle because Flipkart was offering a flat rate for the exchange. In an offline store they could haggle and get better prices if the phone was well maintained.

In response, the team launched the ‘condition assessor’.All a customer now has to do is run a feature on the Flipkart app—the feature checks how well that phone is functioning and throws up a custom rate. Since Flipkart works with companies that buy old phones and electronics, it already knows the exact value of a used device in the liquidation market. Amar says that on day 2 of the 2016 Big Billion Days sale, around 200,000 phones ran the condition assessor. What’s more, the company claims it makes no losses on any of the exchanges.

On the ease of payments side, besides solutions like PhonePe and features like no-cost EMI, Flipkart is also experimenting with features that remove the hassle or friction from making a payment after each purchase.A feature in the pilot stage right now is Buy Now Pay Later. Flipkart leadership declined to confirm its existence but sources say the feature shows up only for select customers. These customers do not have to pay when they purchase a product on Flipkart. The payment for all purchases made in a given month needs to be made only in the subsequent month. And just like with credit cards, there are late payment charges to be paid if they don’t pay up on time.

Underlying all these innovations is data. Flipkart’s data sciences team analyses terabytes of data in real-time, builds dashboards during sales events so the business team can make quick decisions and helps the company understand the rapidly evolving nature of the e-commerce customer. Flipkart’s data team is working on everything from dissecting brand awareness and predicting user response to a product, to understanding user tastes and preferences and detecting fraud.

The data team is helping the product team figure out, what Ram calls the most difficult problem in the world—women’s size. The issue Flipkart, and all other retailers face, is that two women may have similar dimensions but one will find a product, say a kurta, is a good fit and another might not feel the same way. The data team is figuring out why that happens. If they can crack the answer, the algorithm might be worth a whole lot more than a simple fix.Is this good enough?

“Flipkart needs to constantly keep innovating to keep improving customer experience. Only that way can they retain their market share lead,” says Mrigank Gutgutia, e-commerce expert and Engagement Manager at RedSeer Consulting.The team at Flipkart is definitely full of ideas. Flipkart has stated its intention of offering an entire galaxy of apps that fulfil all the shopping needs of a user—from buying groceries to booking flight tickets. Ram says that with technologies like artificial intelligence (AI) and virtual reality (VR) becoming mainstream, predicting everything from style to loan amounts could become automated and more intuitive. Advancements in speech technology too will change the way Indians shop, he says. “I think of a future where no one will need a Flipkart app, they will just speak to something and Flipkart will just take care of their needs, and we can predict what they need too,” says Ram.

The only issue here is the sneaking suspicion that Amazon is already solving these problems. But it is doing so in developed markets. Amazon’s voice control system, Alexa, along with its smart speaker, Echo, allows voice shopping. At the Consumer Electronics Show (CES) in Las Vegas this year, LG announced an Alexa-integrated refrigerator that allows voice-controlled online shopping. Amazon is even trying to change how people shop offline with its Amazon Go store that it is piloting in Seattle. Using computer vision, sensor fusion and deep learning, it begins the tracking process when a customer takes a product off a shelf and charges their Amazon account directly. This means a shopper can just pick up the product and walk out. (At present, only Amazon employees have access to the store.) How much time will it take Amazon to bring these solutions to India? Won’t Flipkart fall behind as it is still in the early stage of developing such AI-based solutions?

This is where India’s diversity, heterogeneity and complexities—often seen as business challenges— could well serve as defences for Flipkart. Amar says when Flipkart talks about conversational search as a challenge, most developed-market techies respond by saying that the West has already found a solution. “It is not the NLP (natural language processing) that is complex here. It is actually the deep learning of the many languages in India. Some are even hybrid languages like Hinglish. This is a problem the West doesn't have to solve, as it is (dealing with) just English,” says Amar.Then there is the Indian customer in all his idiosyncratic glory. In a market with very little standardisation of products and where a vast majority of consumers are not exposed to organised retail, retail is more of an art than a science.

Flipkart’s advantage lies in a little something called ‘Action Shoes’. No, that’s not an error. Flipkart kept getting searches for ‘Action Shoes’ from Lucknow, but the customers weren’t actually looking for shoes of that brand. There were a lot of searches for Action Shoes but very few purchases. So, on one of his road trips, Amar visited Lucknow, went to a shoe store and asked for ‘Action Shoes’. Out came sneakers of all brands. That’s when it hit him—folks in Lucknow call sneakers ‘Action Shoes’—a shoe brand had become a generic term! “This is not a problem the West had to solve,” he remarks.

While this poses a big challenge for Flipkart, the company knows it is just as big a problem for Amazon. This levels the field. This gives them hope. And makes them audacious.