Flipkart vs. Amazon

Image Credit: Sanjay Piplani

Forget the furious debates on discounts, GMV, and valuation. It was always about code.

PROLOGUE
Late 2014.
At a small hotel in Bengaluru, the three Bansals of Flipkart—Sachin, Binny, and Mukesh—were sitting with the team of subsidiary fashion e-retailer Myntra for a strategy offsite. Sachin conducted the proceedings with the air of a father figure, a senior manager recalls. Mukesh played sutradhaar. Binny summed up the discussion points and the next steps. Sachin wanted to press on with an aggressive mobile strategy that he believed held the key to the future for the Flipkart-Myntra combine. His inspiration came from consumer behaviour. As the head of a company that derived a large share of its revenue selling phones, he knew the age of the smartphone was here.
In 2014, smartphone sales in India touched 140 million. Mary Meeker’s Internet trends report, considered a bible for the digital economy, says more than 40% of online sales that year was on smartphones. “Consumers are showing us the way,” Sachin goaded the team. “If we don’t focus on mobile as the primary channel, somebody else will.” There was another, more urgent reason to move rapidly on mobile. Amazon, the planet’s oldest and biggest online retailer, was charging ahead in India. Flipkart couldn’t afford to let it gain ground on mobile. If the average young smartphone user has space for five apps, analytics firm App Annie typically names WhatsApp (messaging), UC Browser (browsing), Facebook (social media), and Hotstar (entertainment) or TrueCaller (caller ID) as four of them.
If there’s space for just one retail app, Sachin could not let Amazon seize that. At stake was something far more precious than revenues: user data, the holy grail for e-commerce firms. It is this data that drives engineering and innovation—the zillion lines of code that lure customers into buying more, or revealing their preferences so that they can be shown things they may want to buy. Whoever controls this data controls the market.
Amazon had lost China, bullied by local players who understood the market much better. Jeff Bezos, its talismanic founder, was prepared to do whatever it takes to win in India—including signing a $2 billion (Rs. 12,634 crore) cheque meant to outspend anybody who came in the way. For instance, on the counting day after the 2014 elections, when Indians went online in unprecedented numbers to catch result updates, Amazon India’s digital advertising spend equalled what it then spent in a month. The danger for Flipkart was that if Amazon managed to push its mobile app through a similar blitzkrieg, it would nullify Flipkart’s biggest competitive advantage: deep knowledge and granular data on the local user, based on its seven-year head start.

Amit Agarwal, vice president and country manager, Amazon India

Without that advantage, it was a matter of time before Amazon trounced Flipkart in its own backyard. There was only one road ahead: Focus every last engineering dollar on building a wall around mobile.
From early 2015, Flipkart started herding its mobile site users to its app, shutting down phone-browser traffic. Myntra went a step further—it shut down its desktop site and became app-only. Sachin knew he risked losing a large chunk of users. But it didn’t worry him. Rather, he saw it as a way to identify a mass of loyal users on mobile. Their inputs and spending habits would power Flipkart’s innovation engine and give it the edge it needed to defend its turf against Amazon.
In hindsight, the thinking proved too abstruse for his own team. “Sachin was committing the company to a future nobody else could see,” recalls the former senior manager at Myntra, who witnessed the then CEO’s audacious thinking. For Sachin, though, audacity wasn’t a new thing.
Subrata Mitra of Accel Partners India, Flipkart’s first investor, spotted the trait in Flipkart’s formative years, when the young company punted on cash on delivery (CoD) in a country where the online buyer was still a fuzzy, none-too-trustworthy entity. “They met their targets in half the time, and asked: ‘can we go to the next step?’ Sachin would say 10x, even though the board would [say] 2x or 3x … It was fascinating—and scary,” Mitra had told me during a chat.
Audacity is essential to build a company that claims to engage with 75 million users and deliver 8 million shipments every month—scarcely believable if you consider India’s total active Internet user base was only 32 million when Flipkart began in late 2007 as a startup delivering books. It was the same audacity, unforeseen in India’s Internet businesses, that would prompt another investor to tell the CEO of a marquee digital company: “If these guys can meet their numbers, I will always be four quarters ahead of competition.”
That investor, Lee Fixel of Tiger Global, issued his first cheque of $10 million to Flipkart in 2010, en route to becoming its biggest investor. Sure enough, Flipkart came to dominate India’s e-commerce market. Its two biggest rivals, Snapdeal and Amazon India, were roughly half way behind as of March 2015.
Also read: Trail of the Tiger
But as Flipkart ran its race, something sneaked up on it. Audacity suddenly seemed to be losing out to the other stuff that decides winners: discounts, gross merchandise value (GMV), valuation. Particularly with Amazon breathing down its neck and funding drying up, Flipkart became a meme in furious “future-of-Indian-e-commerce” debates.
This is the story of what those debates forgot.
Back in 2014, when Flipkart was announcing its acquisition of Myntra, I had asked Myntra’s co-founder and then CEO Mukesh Bansal to name the first thing about the acquiring company that caught his eye. “I admire Flipkart’s very deep investment in technology,” he replied. “Not many Indian companies can claim that.” As an example, he spoke of Flipkart’s strides in search engine optimisation (SEO). “Flipkart really cracked SEO, way back in 2008,” Mukesh said. “We had a six- to nine-month head start [in funding] over Flipkart. But suddenly, Flipkart’s traffic was growing every day.” Flipkart eventually overtook Myntra. “We thought they had no money but were very impressed with the innovation and thinking behind leveraging SEO, which no one in the ecosystem knew at the time.”
Flipkart wanted to be the reason Indians bought products on the Internet. Its focus on technology to solve product e-commerce for the domestic market put it in a league of its own: Even the storied Indian IT and BPO industry derived nearly 90% of its profitable revenues from global enterprise clients.
In contrast, Flipkart’s growing battalion of coders—just over 1,300 at last count—worked to solve problems for desi customers—you and me.
“Flipkart started with that technology engine,” says a Delhi-based strategic consultant who has worked with the company. “The founders are technology guys—for them, it is not just an expenditure they incurred in the initial phase. Their mandate to the team is to innovate constantly.” An investor who had turned down Flipkart in 2007 says: “Flipkart took the technology playbook out of Amazon, and built e-commerce for India out of it.”

Binny and Sachin are technology guys. For them, it is not just an expenditure they incurred in the initial phase.

It’s that technology DNA, backed by a staggering execution record, which saw Flipkart hog $3.2 billion of the roughly $6 billion in foreign investments on homegrown e-commerce players, including the trifecta of rival unicorns Snapdeal, ShopClues, and Paytm. If Amazon engineers grudgingly respect Flipkart, it is because of the latter’s astute reading of how to make tech work for India. And nearly three years after Amazon launched in India, it is technology that is the theatre for a battle for the ages: Both Sachin Bansal and Amit Agarwal, country manager of Amazon India, are techies before being CEOs, meaning they have built organisations where tech thinking has primacy over everything else.
More crucially, it was the choices Flipkart made around technology—chief among them that decision to go mobile-only—which defined the derring-do and exacerbated the fault lines within the company, prompting naysayers to all but write it off. And yet, a clear understanding of those choices, sans drama and intrigue, has been missing from the narrative.
Distilled to its core, this is a classic “homegrown pioneer vs. giant multinational” story—think Nirma vs. Hindustan Lever, Thums Up vs. Coca-Cola, or Mahindra & Mahindra vs. Toyota Motors—with technology as the leitmotif. Flipkart has scale and local nous. Amazon has staying power and a platform it has seasoned globally for 21 years. Which combination will carry the day?
The Flipkart founders did not speak to us for this story, but Amit Agarwal did. We also managed to speak to several Flipkart insiders and players in the wider e-commerce ecosystem who had a ringside view of the company’s tech playbook and the thinking behind its thrust to mobile. Many of them didn’t want to be named, fearing they will be dragged into the increasingly sensationalised debate. This story also gives them a voice.
I
THIS STORY COULD HAVE SEVERAL BEGINNINGS, but none more telling than a development from last November. Amazon became the second-largest U.S. stock holding of Tiger Global, one of the world’s most-watched tech investors whose every move is thought to be an indication of which way the wind is blowing. In India, this was the time when all the talk was about jittery investors pulling back from underperforming copycats of Valley successes. Business daily Mint said it with a revealing headline: “Flipkart’s largest investor Tiger Global raises stake in Amazon.”
This January, Amazon’s chief financial officer Brian Olsavsky claimed Amazon India’s gross sales in the December 2015 quarter surpassed its full-year sales of 2014. It coincided with growing chatter about how Flipkart had missed its GMV target for the full year.

“Three years ago, Flipkart was the undisputed leader and customers swore by it. Today, people say: Amazon gives the best experience,” says an industry source in Mumbai. “Flipkart is still a respected brand, but for it to command the valuation (north of $15 billion before the recent markdowns), it has to answer one question: Who am I?”
The mobile-only bet was to be the answer to that question.
Flipkart wasn’t the only one pushing mobile. In April 2015, Snapdeal bought out mobile recharge company FreeCharge for $400 million. By August, it had got Amitava Ghosh, former CTO of cab aggregator TaxiForSure, to head its Bengaluru development centre of 250 techies. Another e-commerce upstart Paytm was already leveraging its origin as a mobile wallet company to differentiate itself. Even in Amazon’s plans mobile had a disproportionate share.
“As a percentage of traffic, mobile continues to be bigger [than any other medium],” says Akshay Sahi, who heads customer experience at Amazon Seller Services. “When we innovate in India, up to 97% of the product work is for mobile. Everything is focussed on the app. That is the flagship experience we are building.”
But as the leader of the pack, Flipkart’s all-or-nothing plan generated the most intense commentary. (Some of it was borderline bizarre. For instance, serial entrepreneur Alok Kejriwal described the desktop shutdown as a ploy by Flipkart to slow down sales, which would help the company in pruning losses since each sale still costs the company money. Sachin Bansal replied with a widely circulated takedown of Kerjiwal’s theory on Twitter.)
As it turned out, the plan was a non-starter. In November 2015, seven months after it all began, Flipkart revived its mobile site amid a barrage of complaints from angry customers. Myntra would also reopen its catalogue on the Web and is working to restore its desktop site too. But that one move gave Amazon just the opening it needed to go in for the kill.
Sachin Bansal’s two biggest calls before this—building in-house logistics and delivery and CoD—had become industry standards. But with this gambit, he practically sacrificed the desktop and mobile browser users of Myntra (a net 15% drop-off from its growing aggregate user base) as well as some 7% of Flipkart’s mobile users.
Inevitably, there were tensions within Flipkart. “What people see from the outside is how consumers left when the mobile strategy backfired. But even internally, it created a lot of strife,” the industry source says. You can see why, if you consider that Flipkart had spent millions of dollars to amass these users in the first place. Letting them go gave Amazon an unexpected leg-up, and it wasn’t going to squander it.
II
USERS ARE ONLY one part of this battle. The other, less visible part, involves sellers. Winning sellers is critical to dominating e-commerce in India. Here, government rules allow 100% FDI only if you follow the marketplace model, where sellers or merchants supply goods rather than e-commerce companies peddling their own inventory. The rules also cap sales per merchant on any marketplace at 25% of total sales.

Akshay Sahi, head, customer experience, Amazon Seller Services

Amazon went after sellers right from its entry in India and has scaled its network from 100 in 2013 to 85,000. Flipkart, on the other hand, had to adjust its focus to keep pace—since its inception, when there was no regulatory dicta for e-commerce, it had focussed all its resources on getting more customers and not sellers. (Flipkart breached the 100,000 seller mark this February.)
“You can’t rely on integration API (application program interface, a tool that makes it possible to manage software applications) or sophisticated mechanisms in India because these are typically small [shopkeepers],” says Amazon India’s Agarwal. A video by the company captures how shopkeepers generally do their stock checks. They scribble code letters to help them remember if a product is in stock. The role of technology is to replicate the comfort of this unique physical behaviour on an app. Only that will hook sellers to the app and keep them coming back, says Agarwal.
“When sellers scale up their business on a platform, they get access to more powerful technology, which helps boost exposure of their inventory,” explains a partner who works on Flipkart’s Sellers Hub. But scaling calls for a lot of handholding, since most sellers are exploring the online medium, and the technology driving it, for the first time.
According to a source who works in the seller ecosystem, this is where Amazon scores big. It woos sellers with rich data analytics that can be understood by any layperson. It also invests heavily in smooth resolution of payments. “Amazon reconciles money order to order. In Flipkart, it is not like that,” says a tech entrepreneur who works with sellers to boost sales on both Amazon and Flipkart.
Amazon’s charm offensive also includes the Chai Cart and Tatkal initiatives. Chai Carts are three-wheeled vehicles that serve tea, water, and lemon juice to small business owners even as Amazon reps teach them about selling online. With Tatkal, a studio-on-wheels, Amazon helps sellers launch services in less than an hour, including registration, imaging, and cataloguing services, as well as basic seller training. Both became popular enough to merit a mention in Jeff Bezos’s 2015 annual letter to shareholders.
Then there’s Seller Flex, a feature that allows a shopkeeper to convert his or her store or warehouse into a fulfilment centre by using Amazon software. “It is meant to help sellers who are unable to ship their items to Amazon’s warehouse because shipping costs are too high,” explains Agarwal. In the past six months, Amazon India has launched 21 such nodes. It is an India-first innovation for the company.
The tech entrepreneur who has worked with both companies outlines areas where Flipkart has improved. “Its reports for sellers now offer easy visibility of what is in stock, what is selling well, and what is not. It also tracks the sellers’ performance.” All this may sound like hygiene—except Flipkart is building things ground up, unlike Amazon which can borrow from a well-established global template.
But Agarwal brings the conversation back to customers. “If you look at mobile web and PC web traffic, we are the leader by far in terms of time spent as well as conversion. In pretty much all the metrics that customers—not VCs—care about, we are the leader,” he says. He is dismissive of the “battle” hyperbole. “We are only focussed on how we are going to position India as Amazon’s largest geography over the next 10-20 years—period.”
And to think that Agarwal, Sachin, and Binny once worked from the same office, where all three of them honed the beliefs that would take them on their separate journeys.
III
IT IS THE LAST DAY of the financial year 2016. Fittingly, I am spending it with Agarwal at Amazon India’s office in Bengaluru’s World Trade Centre, exchanging notes on possibly the most talked-about business story of the year. I can’t resist asking him: Did you know Sachin and Binny back when all three of you were with Amazon Web Services (AWS)? Agarwal smiles. “Well, it was a small team, and I know the people who work for me.”

Mrinal Chatterjee, director (technology), India Payments,
Amazon Development Centre

Those were anxious days for Agarwal. At 32, he had been asked to lead Amazon Development Centre’s (ADC) India operations. (Amazon has similar units in 13 countries, including China, Japan, and five in Europe.) Since 2004, he had been in charge of about 50 software engineers who were innovating for the Amazon.com platform. He had also been entrusted with launching AWS here, a cloud services outfit that is now Amazon’s most profitable business. But he wasn’t happy.
Agarwal was an early believer of the Amazon story. A Stanford post-grad in computer science, he had joined the company’s Seattle headquarters in early 1999, a year before the dotcom bust, drawn by Jeff Bezos’s 1997 letter to shareholders. That letter has appeared in every Amazon annual report since. “Today, online commerce saves customers money and precious time. Tomorrow, through personalisation, online commerce will accelerate the very process of discovery,” read one of its most prophetic parts. But Agarwal was not assigned to the e-commerce business. It bothered him.
“[My work felt] very tangential. Every day, I was sceptical: What am I doing in this team?” Agarwal recollects. In hindsight, the stint taught him the true meaning of using technology to become the world’s most customer-centric company. Amazon was going after a diverse set of customers: creators, developers (AWS), readers (Kindle), online consumers, sellers. “Working with ADC taught me how flexible our brand and culture is to foster innovation.”
Amazon was foremost a technology company, and Agarwal’s team played a critical role in its evolution by shaping its relationship with developers.
In November 2007, Agarwal moved back to Seattle as “shadow” and technical advisor to Bezos. “We [shadows] had to help Jeff make more effective use of time. I was always travelling with him and sitting in on his meetings. It helped me develop a holistic view of Amazon,” he says.
Many others stayed on at ADC or took up assignments at other Amazon locations worldwide before returning to launch e-commerce in India. But there were two coders who moved on: senior software engineer Sachin Bansal, and Binny Bansal, one of Sachin’s references to AWS.
Like most young coders, they were geeks. Sachin loved gaming, while you were likely to catch a painfully shy Binny reading Stieg Larsson. Their working styles complemented each other. Sachin would set outrageous goals, Binny would lay out the steps to get there. People today call Binny’s approach binary: 0-1, this or that. “He is very sharp in terms of planning for the next level. He will ask incisive questions, and spot where a plan can go wrong,” says a senior manager who has seen them work in tandem. “With Sachin on the other hand, it is about the size of the ambition.”
Sachin quit ADC first, in 2007. He moved to Binny’s apartment, where the duo chiselled the idea of Flipkart by studying the success of India’s online commerce pioneers—ticketing websites like MakeMyTrip. Sachin told Fortune India in late 2011: “There were lots of people buying tickets online, but not many people looking to buy products online. Why should that be? If people can trust Internet companies with their credit card information online for tickets, why not for products?”
He identified customer service as the missing link. “There are significant problems, and if somebody solves those problems, a lot of value can be generated. We jumped in with that intent: bring customer focus into the industry,” Sachin said. Last-mile logistics was to be Flipkart’s big weapon to ensure that buyers who had never shopped on the Internet got a hassle-free shopping experience.
It would be expensive. But Binny argued that technology offsets the cost of building physical presence, after the company crossed the top 30 cities. This—suburban India—was the threshold where access to products as well as authentic customer data typically dwindled in physical retail chains. This was the big opportunity for e-retail.
IV
BY 2011, Flipkart had committed to expanding its logistics and delivery arm, which accounted for a large chunk of its 3,500 employees. Technology was only 5% of this (180 persons). Even that was a big jump given as recently as two years before, the tech team was just Sachin and Binny. Back then, Flipkart used shared servers. But its tech focus was sharply evident. “Everything we do is technology-based,” Sachin told me in late 2011, when it clocked 10 million visits a month. (Today, the number is 10 million a day.)
“The website you see is just 10% of what the technology team does,” Sachin explained. At the backend, Flipkart had large servers. There were payments-tech teams to integrate with the IT systems in banks, Visa, MasterCard, and so on. “The largest tech team we have is in supply chain, which builds automation, systems for warehousing, logistics, procurement, even trip planning and delivery prediction,” Binny weighed in.

Flipkart was then delivering in seven cities. The focus was on reliability—building out logistics and improving warehouse systems and technologies to handle scale. But all along, its engineering team was readying the platform to serve many, many more destinations. For Flipkart customers in 2011, the order actually reached home in the promised time, often in two days in large cities. It was solving a far greater number of problems than early starters eBay India or Fabmall, which had entered the space before the turn of the century, or even newer competitors like Snapdeal and ShopClues.
Also read: Snap Kart Flip Deal
Flipkart was spending much more than any of these companies, yes. But it was also deploying more tech—and faster—to grab the lead in amassing customer data. In 2011, its contemporary Homeshop18 had 2.3 million unique visitors while Naaptol had 2.15 million—Flipkart was ahead at 2.7 million, according to November 2011 comScore data. (Snapdeal was still a couponing site, and ShopClues was  just born.) By June 2013, when Amazon entered the market, Flipkart’s logistics arm was already clocking 90,000 shipments a day.
By February 2014, Flipkart had breached $1 billion in GMV, a term that denotes the value of the inventory that gets sold on an online platform. (It’s a much vilified metric but remains a good indicator of a market where buyers are not naturally inclined to order products online.) But with Amazon at its heels, it had to turn its attention to building a lasting edge. The stage was set for Sachin Bansal’s big mobile bet.
ITS MASSIVE SCALE and implications meant the bet could have only come from Sachin. Ex-employees say Sachin is vital to the product and engineering pack that is core to Flipkart and its culture. “Good entrepreneurs take gutsy calls with incomplete information and live by those calls. Sachin can take big calls and rally the team,” says a person who has been closely associated with Flipkart.

In both big calls of the past—last-mile logistics and CoD—he had played oracle. His goal-setting had a galvanising effect on techies, especially because Indian consumer Internet companies before Flipkart were rarely ambitious about local markets. “Flipkart has the best technology any Indian company could build,” says a former engineer. “And it has to do with Sachin.”
The engineer says that when Flipkart solves a technology problem, it isn’t for the sake of putting out just another function or feature. It runs deeper. If catalogues are to be created for the website, compiling information from tens of thousands of sellers, it involves designing a standard framework for 300,000 and more listings per month—cutting out duplication. “Flipkart won’t be happy with just a catalogue that looks and feels good. The goal is accuracy. You have to win customer trust because that catalogue is all a customer sitting far away can go by.” Techies thrive when coding is done for that greater goal: trust, reliability, solving problems for the local user they can relate to, not a faceless ‘onsite’ client sitting thousands of miles away.
Flipkart had to strengthen another skill—execution—in the face of Amazon’s well-oiled machine. This is what Binny, with his eye for detail, would oversee.
The Myntra acquisition added a formidable third leg to the Sachin-Binny partnership: Mukesh Bansal. Myntra had given Flipkart the largest market share in fashion. It also boosted Flipkart’s performance on a range of vital parameters: frequency of visits, the number of items bought per order, repeat purchase, and higher margins compared to its other staples, books and mobile devices. Mukesh, who had earlier spent six years in Silicon Valley, was appointed to the Flipkart board, in addition to a broader management role as Flipkart’s head of fashion.
Also read: Merger, no. IPO, yes. Mukesh Bansal's plans for Myntra
The Flipkart founders were excited about the new configuration. They often stated in public forums that their investors have allowed both companies to run independently. Flipkart and Myntra had grown from being startups, and the three co-founders enjoyed exchanging learnings.
There was a lot to learn from Mukesh. He had survived an attack from Rocket Internet-backed Jabong and kept Myntra ahead. Even under capital constraints, he had invested in hiring leaders like Ganesh Subramanian from Walmart as chief operating officer and Shamik Sharma from the Bay Area to head technology. Pre-acquisition, Myntra would often lose out to Flipkart in hiring tech talent because it was perceived to be more of a fashion brand than a tech venture. Sharma’s leadership made a difference.

Given that track record, Mukesh was asked to put tech leadership in place at Flipkart. The ideal candidates would have experience in solving problems for 200 million-plus users—think Google, Twitter, LinkedIn, or Square. During our conversation at the time of the Myntra-Flipkart deal, Mukesh laid out the task: “We have an aggressive agenda going forward to grow talent on both sides: in India and globally. In fact, we are seeing a massive shortage of technology talent.”
By mid-2015, he had hired Saikiran Krishnamurthy as COO of Flipkart’s commerce platform and Ananth Narayanan to succeed him in Myntra as CEO (both are ex-McKinsey). Another key hire was chief product officer Punit Soni, who had earlier led a series of important product portfolios at Google. Soni would emerge as a prominent player in the internal strife around the mobile-only strategy, as we’ll see.
Another ex-Google recruit was chief technology officer Peeyush Ranjan, who worked as senior director of engineering for the search giant based out of Mountain View, and had also led engineering at Google India out of Bengaluru. Ranjan understood India. At Google, he had closely overseen the development of Google Map Maker, a product that allowed users to edit Google Maps. It was a solution designed specifically for India, where publicly available addresses are often fragmented and inaccurate.
Ranjan also understood the value of the local coder solving local problems. At Google India, for every one engineer focussing on India, roughly 10 were solving problems for the rest of the world. “The utility of product managers to the local market was not that high, even though they understood what should be done for the locals,” he says. “Now with India becoming important for every company, the same product managers are very valuable.”
By early 2015, Flipkart had a new management structure. It would now have three decision-makers—Sachin, who would head new initiatives, Binny, who would look after supply chain, and Mukesh, who would be responsible for the core commerce platform, with product and engineering (Soni and Ranjan) reporting to him. In hindsight, this was another area where Amazon India had a better plan: It made sure that technology, commercial, and operations all rolled up to one boss—Amit Agarwal—who would in turn execute Bezos’s vision for India.
V
BEZOS BEGAN TYPING his 2010 shareholder letter with the following words: “Random forests, naïve Bayesian estimators, RESTful services, gossip protocols, eventual consistency, data sharding, anti-entropy, Byzantine quorum, erasure coding, vector clocks … walk into certain Amazon meetings, and you may momentarily think you’ve stumbled into a computer-science lecture.”

Peeyush Ranjan, chief technology officer, Flipkart

“The e-commerce platform is made of hundreds of software services that work in concert to deliver functionality ranging from recomv  mendations to order fulfilment to inventory tracking,” Bezos wrote. Two years after the financial meltdown, one thing had to be made clear to financial investors: The geek will inherit the earth.
Amazon’s balance sheet tells the story: “technology and content” is its third-highest operating expense that cost it $12.5 billion globally in the calendar year (CY) ending December 2015, after “cost of sales” ($72 billion) and “fulfilment” ($13 billion). More interestingly, technology accounts for the highest stock-based compensation ($1.2 billion in CY15). Of course, Amazon’s technology and content expenses don’t reflect e-commerce alone. They include other businesses like AWS and the on-demand video network that is taking on Netflix. But with e-commerce as its flagship, Bezos knows technology spending is vital for long-term dominance.
Why does that matter to Amazon India? It does because Amazon’s tech teams are united globally on one platform. The ADC Agarwal incepted in Bengaluru as a 50-member team in 2004 now employs more than 10,000 across four Indian cities. “When we build software technologies, it has to handle Amazon’s [global] scale—not just Amazon India,” says Mrinal Chatterjee, director for payments at Amazon Development Centre. “It ensures we don’t have to rework the same technology if it is required in another market.” This is a machine at play, with systems and processes honed over many years.
Then, there is the exchange of ideas. Amazon India can afford to prioritise mobile product development, as the other global centres continue work on the desktop product. In its home market, the U.S., 78% of Amazon’s traffic still comes from the desktop. There’s also project-specific collaboration. For instance, the work in India on two-factor authentication for payments draws from, and contributes to, the global team’s work on multi-factor authentication which came into effect in Europe from the start of this year. A delegation of Amazonians from Europe was in India in March to understand two-factor authentication.
New recruits at Amazon undergo engineering boot camps and hackathons across geographies. Chatterjee has made passport mandatory for his team members after they complete two years. “If engineering teams are working in isolation, we will not get the best. We have to actively collaborate,” he says.
Amazon India’s Sahi says: “As you build architectural solutions, you run into errors and challenges.” Having a global technology platform helps because “our folks have insights from others who have worked on those platforms, telling them ‘this is going to work’ or ‘this is going to break.’”
For an Amazonian in India, tasks on the collaborative platform broadly come under three buckets. First, engineers sitting in another location can help develop a solution for India. Second, there may be features Amazon India is working on which are already on the global product road map and need to be accelerated. In this case, engineers from India step in. And third, building a solution for India that is not on the global road map. “We build such solutions and sell them to the rest of the world,” Sahi says.

Sahi cites an India-first feature exported globally: accounts on the Amazon platform for first-time Internet users who don’t have e-mail accounts and operate only on mobile. When the user opens the Amazon app, it automatically reads the mobile number in the background. The user is prompted to set a password based on an OTP (one-time password), which too the app can read automatically. “This ends up removing a number of steps. When you think of the latencies involved, all of it adds up,” says Sahi. “We have done a bunch of other experiments—some successful, some not—but users don’t find out about them.” It is a culture of iteration Amazon has perfected over two decades.
Sachin was only too familiar with Amazon’s technology prowess. As someone who had always set Flipkart’s vision, he knew he had to make a big, bold statement to keep Amazon at bay. But even he may not have fully understood how the consequences of his decision would ripple through the company he had built.
VI
THE ROLLBACK of the mobile-only plan wasn’t Flipkart’s biggest failing. App Annie says it is still India’s No. 1 shopping app, with downloads breaching the 50 million mark this February. (According to an App Annie analyst, Amazon led Flipkart on Google Play downloads in Q1 2016, specifically in January and February.) What disturbed Flipkart’s cheerleaders was its inability to coalesce as a team around the plan, or rally around each other to modify it, or contain damage and win the perception battle when things unravelled.
“Sachin’s strategy was brilliant. If anything, we couldn’t execute it as fast as we would have liked,” says a senior manager. To be sure, it wasn’t an impulsive call. He had mulled over it for almost three quarters—analysing every possible data point and leading indicator—and only then did he put it forward as a directional shift, says another manager.

Sachin’s rationale was this: Going app-only would mean users would have access to only one channel. The company would be able to capture the shoppers’ experience and what they were searching for, in a manner that it simply couldn’t from a browser. Even if users uninstalled the app, Flipkart’s coders would take away valuable data on device storage and functionality limitations, which would help them innovate for future users. The larger question: “If you are indexing big on mobile, how do you deliver discovery without taxing the customer?” as Ravi Garikipati, a senior vice-president at Flipkart, puts it. In effect, the company could innovate based on actual learnings of the limitations users experienced
But the team wasn’t prepared for all the complexities of what Sachin demanded of them. “What is the most frictionless way for users to discover your platform? It is still the mobile browser,” says a source closely associated with the company. Flipkart shut off that option for seven months.
Other doubts swirled. For instance, how many apps would people keep? “If you’re really looking for the next 100 million users, you have to assume they don’t have enough phone memory to store more than five apps,” says the Flipkart associate. Why push them away by cutting out the browser? “It was about priorities,” he says. “Sachin wanted the thinking in the team to be very focussed. But there was probably no need to shut down mobile web.”
The confusion within the ranks claimed a series of high-profile scalps. On the evening of 10 February, Myntra’s employees—past and present—discovered from a WhatsApp update that Mukesh had stepped down from Flipkart in less than a year. Given his rapid rise in Flipkart, the news shocked many. Conspiracy theories started flowing, the most common one suggesting a fallout with Sachin and Binny. That’s hard to prove. But in a conversation with one ex-employee, I hear something that may explain a fallout, had there been one: “Flipkart is a very different company from the kind that Mukesh built at Myntra.”
When the Flipkart founders and board offered Mukesh the role of CEO of commerce platform, it was taken as a sign of the company’s maturity. Vivek Wadhwa, director of research at Duke University, has a keen eye on Indian tech entrepreneurs. In March 2015, in The Economic Times, Wadhwa compared the development to Google co-founder Larry Page allowing Eric Schmidt to take over as CEO under pressure from the board. Sachin wasn’t under such duress; Mukesh had the board’s trust to free up Sachin from the nitty-gritty of operations, so that he could dedicate all his time on big-picture ideas and the defence against Amazon.
“When it comes to performance, Mukesh is extremely sharp. He sets very high bars. Many a time, people don’t get him,” says a former colleague. One of the tougher challenges Mukesh undertook was to make Flipkart a leaner, more efficient organisation. Its headcount was ballooning every quarter. Its legal team alone is estimated to be around 40 persons (almost as big as a mid-sized legal firm). The HR department is some 400 persons.
By the end of 2015, Flipkart was a 35,000-person organisation; almost half of them were part of the supply chain overseen by Binny and around 15,000 were on contract. Mukesh was heading the core commerce platform, with fewer than 2,000 employees, including engineering and product. He wanted to make this unit leaner. In the 18 months Mukesh spent at Flipkart, he “wanted to change it from being a people-dependent organisation,” says a headhunter in the know. But the mobile-only strategy would prove to be the biggest distraction for the core team.
According to three ex-employees of Flipkart, Soni, one of the company’s star hires, was opposed to the idea from the get-go. As head of product, his task was to build customer-facing features. Turning off users went against that very basic tenet, especially if Flipkart was serious about building a following of 200 million shoppers. One industry observer comments: “As the breakout leader, Flipkart had the option to build on its strengths—its  user base—and out-execute Amazon. The app-only move was a big strategic blunder,” and showed an unhealthy obsession with competition. Soni flagged off his concerns with Sachin in early 2015 but couldn’t change his mind.
Both were right in their own ways. Soni, responsible for product, wanted to expand the existing user base and keep it user-friendly. The mobile strategy would estrange users and sabotage the mission. The outcry on social media to Myntra’s app-only strategy was vindication for his position. A strategic consultant in Bengaluru tells me: “It is the equivalent of shutting two entry points of a mall that has high footfalls, and telling buyers where to enter from.” To the outside world, it smacked of arrogance.
It is not clear how Sachin countered Soni. But two sources try to paint a picture of what he might have said: There are fewer than 40 million online buyers of products today. If Flipkart had to build a strategic advantage, it had to innovate for users beyond the 50 million mark, which is a kind of industry benchmark. Reaching 200 million users, its ultimate goal, entailed going really deep into India, dominated by lower-end smartphones with less storage. Flipkart needed to aggressively pursue this path before Amazon got there.
It was an irreconcilable rift. Soni quit.
According to a former senior manager at Myntra, Mukesh was initially in favour of taking Myntra app-only. But here’s the other thing about him: “With better information on the table, Mukesh can change his decision.”

The numbers confirmed Amazon’s growth after Myntra vacated space. Amazon claimed in a press release that in tier I cities, it saw an almost 2.5x increase in apparel sales in terms of ordered units between January 2015 and January 2016. It also saw a 4.8x increase in ordered units from tier II cities. All these cities clocked a growth of 3x or more in comparison to January last year.
Myntra’s business team was expecting a drop-off in customers. At first, the magnitude of it (15%) did not worry Mukesh, according to two managers at Myntra. The app was drawing users to offset those who went away. By January 2016, the app had 8 million active users—almost twice the number 12 months before. But Amazon was beating Myntra’s discounts, was more visible on TV, and spending aggressively on digital marketing to bag new users, on both desktop and mobile. Two industry sources dispute this, saying Myntra was also spending big on discounts.
In the October-December quarter of 2015, Myntra ranked fifth among shopping apps (by monthly active users) on App Annie, behind Flipkart, Amazon, Paytm, and Snapdeal. But according to a source, with more and more data proving Amazon’s surge, Mukesh grew uncomfortable. With the desktop still live, there were fewer user dropouts for Flipkart compared with Myntra. But the ones who left were turning to Amazon. There were talks that Flipkart may also go app-only. As chief of commerce, Mukesh was responsible for Flipkart’s overall business performance.
According to the source, by now he was fundamentally opposed to the experiment because he believed it was not customer-friendly. But Sachin stayed firm because this was a strategic call. Around three months later, there was a management rejig at Flipkart. Binny was made CEO, and Sachin became executive chairman. Mukesh moved on.
Fortune India gathered a variety of market-share data from industry sources. A Morgan Stanley report in February 2016 puts Flipkart’s share at 45%, which few in the industry agree with. The consensus is: Of the roughly $12 billion GMV that the five e-commerce companies fought for in 2015, Amazon has taken its share to above 30%, while Flipkart has slipped below 40%. One analyst estimates that Amazon is more than 70% of Flipkart at the end of fiscal year 2016. This means if Flipkart sold goods worth Rs 100 on its platforms, Amazon sold goods worth Rs 70 in the same period.

Ravi Garikipati, senior vice president, Flipkart

Did Flipkart let go of the four-quarter advantage that Lee Fixel envisaged early on? Yes. Did the lower GMV outlook dent its chances of raising big-ticket funds? Yes, because fundraising is often a multiple of market share. You could also say that Flipkart gave away its hard-earned momentum to Amazon. But there is one thing Flipkart got right: In the face of the most daunting competitor in the world, it leaned towards innovation and invested in a world-class technology leadership team.
For a moment, think about Big Retail in India. The industry is now riding the omni-channel wave, bullish on the imminent rise of the 4G-enabled shopper. But has Future Retail or Reliance Retail faced the likes of Walmart, Target, or Carrefour? No. Flipkart did face its biggest challenger—and it responded in the only way it could, by making bold calls based on technology.
Sachin’s push may have been too aggressive, but it saw Flipkart’s engineering team recast itself as a mobile-first outfit. If it hadn’t, its scale may have become a burden, as had happened with Nokia’s feature-phone business in India when Android and Samsung disrupted it. Or, think of how Meru Cabs, stuck in its desktop legacy, was defeated by Ola, which was quickly engineered for the app generation.
Ultimately, if you cannot trust technology and take quantum bets, there is no long-term future for you in e-commerce. Ask Bezos, who in the 2016 shareholder letter puts “eagerness to invent and pioneer” high up on his to-do list, but after “customer obsession”.
EPILOGUE
Last November, Flipkart launched Lite with the aim to deliver an app-like experience on the mobile browser. If you are a Lite user, you could think of it is a 4 kb version of flipkart.com. “When you send people from a website to an app, the conversion is 2% to 6%. We got 83% conversion to Flipkart Lite,” says Ranjan, Flipkart’s CTO. He led a team of five engineers who developed Lite in 42 days.
The Internet in India has an anaemic average speed of 2.8 megabytes per second (Mbps)—among the worst in the world (see graphic). Ranjan was largely able to move the needle at Flipkart from being mobile-only to mobile-first in keeping with this sobering reality. “There is a lot of thinking we went through. Our ultimate goal is not to get the user to install an app. We want to give the user the best experience possible. Sure, this is possible on apps, but can we try another way?”
Ranjan called some friends at Google Chrome for help. Google implemented features like a home screen icon for Lite, which gave it the same look and feel as an app.
“What we realised is, we have to do what is right for our customers. No strategy is going to live longer than that,” says Ranjan. “What products does India need, and how do I build them? How do I solve the hard problems of India through technology? No Indian company or multinational was dedicated fully to that.” In Indian e-commerce, it is still Day One.