Marketing Case: Bajaj Finance

Unit 3, still Professor Das Narayandas's Market-Facing Functions. This time the reading is Bajaj Finance: Building an Omnipresent Financial Services Firm.

Olay was a brand taking itself apart. Belden was reassembling parts into solutions while profits were still rising. Bajaj Finance has already done the assembling — more than fifty products, more than fifty million customers; the case calls it India's largest consumer finance company. In August 2021, it has to decide what comes next: dig deeper into its existing customers' wallets, or race Amazon to open shop in India's smaller towns.

I wrote this before class. I'm putting my judgement down here first, and I'll come back to check it against the discussion afterwards.

Here goes — enjoy:

Case Summary

August 2021, Pune, a stormy day. Sanjiv Bajaj (HBS MBA 1997) and managing director Rajeev Jain are sitting in a conference room. Bajaj Finance — BFL from here on — has US$21.2 billion in assets under management and 50.5 million customers. They want to turn a company that sells loans through stores into one that is "omnipresent": loans, services, and the consumer durables, insurance and investments embedded in them, with online and offline forming a single journey.

There are two camps in the room. One wants to go deeper with existing customers. The processes and people are already in place, and the capital expenditure is modest. The other wants to go to smaller towns. Bajaj's worry is this: the middle class there is growing, and neither brands nor consumer finance have arrived yet. If they move a step too slowly, Amazon will leapfrog them with a financial services platform and catch those hundreds of millions of people. Amazon has already invested US$6.5 billion in India, has 140 million users, and plans to add another 100 million from small towns. It calls itself "Earth's most customer-centric company."

The story begins in 1987. Back then it was called Bajaj Auto Finance, a subsidiary of Bajaj Auto. The Bajaj Group's flagship makes two-wheelers; its founder, Jamnalal Bajaj, was Gandhi's adopted son. At first it only lent to people buying Bajaj vehicles.

After India opened its economy in 1991, global goods poured in. The new middle class wanted to buy them, but inflation meant they couldn't afford them all. There was no credit bureau, and banks only lent to people who could show proof of income. Salaried workers had it; the huge number of people in informal employment did not. Companies like GE Countrywide filled the gap: a small interest-free loan of about US$350 to buy a fridge, a TV, an air conditioner. On the day, you paid a small processing fee and a token down payment, and cleared the rest over 8 to 12 months. Retailers got their money within one to two weeks. Manufacturers were willing to pay a subvention of 5% to 6% of the loan amount, reckoning it was cheaper than discounting all year round.

BAF followed suit and spread to more than 500 towns. Every branch ran its own system and sent its data back to headquarters on CDs once a month, with the paper files couriered separately for data entry. Later, competitors started sharing the subvention with retailers; BAF's margins were squeezed thin, and it had no choice but to share too.

In 2007, the business was spun off, with Sanjiv Bajaj in charge, and in 2010 it was formally named BFL. He had little finance experience. Beside him was his uncle Nanoo Pamnani, who had just stepped down as chairman of Citibank Asia, and they brought in Jain, who had done consumer finance at GE, Amex and AIG.

Their first bet was consumer durables loans. As Devang Mody later put it, this was the only category where you could acquire millions of customers without spending a single dollar — riding on footfall that retailers and manufacturers were already paying for.

But at the time the business was losing money. Competitors approved loans the same day; BFL took 3 to 5 days. The loans were small and short-term, customer acquisition costs were high, and the subvention still had to be shared with retailers. The company had no unified software; information sat in silos, and credit risk was impossible to keep an eye on. Non-performing assets reached 15%, three times the industry level. Nobody in the group wanted to do this thin-margin business. Lending against Bajaj vehicles was far easier.

They did it anyway, with a project called WOW that brought approval time down to 3 hours. In Bajaj's own words:

If parents can tell their children that the TV they bought today will be installed the same day, we create enormous delight, and word of mouth will carry itself forward. Change the experience, and you change the boundaries of the game.

Field credit officers reviewed documents in the store and approved on the spot. Alternatively, store staff phoned a call centre and a rules engine decided yes or no. By March 2008 they were doing 621,000 consumer durables loans a year through 19,000 dealers.

Then the financial crisis hit. Net NPAs rose to 7% of assets. They had neither the technology nor the people for collections. Peers exited in droves. Bajaj, Jain and Pamnani did not. They shut 330 locations, dropped 14,500 dealers, laid off several hundred people, stopped sharing the subvention with retailers, and ended personal computer loans — some people took the computer, sold it on the black market, and vanished. Mody remembers Pamnani saying that if things didn't improve, the business would be shut down.

Only once the books were clean did they start building a different kind of company. They were just getting started; they couldn't afford a price war, and they couldn't win on scale either. As Bajaj put it, no insurance policy or loan contract can be so beautiful that people will pay 10% more for it. Non-banks can't take current-account deposits, so their funding is expensive. So they took a different road: draw on the trust built up by Bajaj scooters and motorcycles, offer the middle class a full suite from loans to protection to investment, and lead with technology. First use data to establish a person's credit, then sell them the next thing that suits them, building a 360-degree relationship.

Very few non-banks were doing this. Tata's finance company was happy to lend to its own suppliers; Chola was content with truck financing. BFL was the smallest horse in the group stable, so whatever it did didn't much matter, and with Pamnani there, the board was at ease. There were two rules. One: build a Mercedes, not a Ferrari, and not a Ford — every new line of business had to be big enough to be worth doing properly. The other was Pamnani's: try anything, but don't develop hobbies.

The first new businesses were unsecured personal loans and small business loans. Without a banking licence they couldn't issue credit cards, so this was the substitute. Later came the EMI card: a pre-approved limit for existing customers that turned into a loan with one swipe at thousands of outlets, starting from about 5,000 rupees. The product range gradually grew to more than 50. Jain said they would take care of retail and SME customers' needs from birth to death, except current accounts and car loans.

They were expanding geographically too. Branches in the big cities, franchises in small places; franchisees shared sales, marketing and service costs on a variable basis, so pushing into ever smaller locations wouldn't immediately wreck breakeven. For big retailers stocking up for festivals, they offered trade advances that could be offset. And there was a famous early scheme: cutting the down payment from the usual 33% of the product price to 1 rupee. At one large retailer in South India, 2,000 people showed up on the first weekend. Manufacturers such as LG put BFL's name in their own advertising. People who came in for a 32-inch TV walked out with a 43-inch; people who came to look at a single-door fridge left with a double-door.

The customer strategy was clear. About 30% of each year's new customers were getting credit for the first time, mostly in smaller, newer markets — places where nobody else was yet. Buying a phone or an appliance in a store, you could get a zero-interest loan within minutes on basic documents. The case notes that even people who could afford to pay in full didn't mind taking the zero-interest loan. Alongside the loan came an extended warranty, which Panchal likened to jam on bread.

Once customers were in, data drove the cross-selling. In Jain's words:

Cross-selling is the holy grail. The dumbest thing an ROE-focused company can do is pay twice for the same customer.

For most people, the second product was the EMI card. For those who had only ever been able to borrow from informal moneylenders, it was often their only card.

Only technology could carry the complexity. Partners supplied the software and hardware — what Bajaj called the arms and legs; the processes and approval rules were written in-house — what he called the brain. A cloud-based loan origination system adapted from a CRM had the rules built in. Retailers were initially reluctant to provide their own computers and internet connections, but once they found that waiting times were shorter and customers less likely to change their minds, more than 3,000 signed up within a year. On Diwali itself, 300,000 applications would come in; if the point-of-sale system went down for 5 minutes, the internal portal lit up red.

In 2015, Jain required that data from every system and business be pulled into a single enterprise data warehouse. Handing over customer data was seen by some as handing over the crown jewels. The resistance wasn't loud — management was either new or already knew the old way didn't work. Those who said "our market is different" were invited to go work for a competitor; Mody says nobody left. Data was aggregated daily rather than monthly, structured data into the warehouse, unstructured into a data lake, stitched together into a 360-degree view of each customer.

Sharing didn't run on slogans. A business unit using another unit's customer data had to pay the source an internal transfer fee. Consumer durables and mobile financing contributed 90% of the customers, and 35% of their own profit came from these fees.

Organisationally, each product was a business unit run by a mini-CEO, accountable for profit, sales, risk and people; only technology and operations had to be shared. New businesses had to double every year. Mature businesses had to grow at least 20%; if they couldn't, they were fixed, and if that failed, shut down. Infrastructure lending was shut down, because whether customers could repay depended on when the government released funds — nothing to do with the core capability of "financing an individual's discretionary consumption." Field and call-centre life insurance sales were stopped too.

Operations were turned into a science. Panchal says they had 10,000 metrics in all. Every Tuesday at 8 a.m., Pamnani would arrive in Pune from Mumbai, and anyone walking in at 8:01 would run into his temper. Risk was separated from day-to-day approvals, and collections were also held accountable for losses. Customers with a home loan defaulted at roughly a quarter of the rate of everyone else. A retailer's approval rate adjusted automatically according to how its past customers had defaulted, and it was made public — so dealers, wanting bigger and faster approvals, would bring better customers to the counter on their own. When a cheque bounced, they got in touch immediately, while the customer still felt a little embarrassed.

Processes were being trimmed, too. Processing a 10 million rupee SME loan secured against property once cost 45,000 rupees, about US$610. Jain asked for it to be halved; the team cut it to 8,000 rupees, about US$110. On the consumer loan side, the cost came down from 1,200 rupees to 800.

By 2019, the report card looked great, and the weather was changing. In rupee terms, over fourteen years AUM grew at a compound rate of 37%, profit after tax at 51%, and operating expenses at 29%; expenses as a share of net interest income fell from 58% to around 31%. Net NPAs were brought down to 0.12% in FY2012; the overdue threshold for classification was later tightened from 6 months to 3, and in FY2021 the figure was still 0.75%. In Exhibit 8, the net interest margin is 13.9%, against roughly 4% for banks. Market capitalisation was US$42.4 billion, with loans outstanding of about US$15.8 billion.

The competition had changed. Banks, with their low-cost funds, were offering instant loans and zero-interest instalments on e-commerce platforms. More than two thousand fintechs had raised venture capital. Paytm had spread from payments into other financial services. Amazon was India's second-largest e-commerce player, with a 36% share, 140 million monthly active users, and 2 million Pay Later customers. The buy-now-pay-later market was expected to reach US$40 billion by 2026.

Small towns were getting richer. Between 2016 and 2025, the number of affluent households with annual income above US$15,400 in emerging cities of under a million people was projected to grow 2.5 times, versus 2 times in the metros. 40% of phone purchases were unplanned. By 2023, 63% of consumer durables sales were expected to be digitally influenced. When buying these things online, 90% of people preferred aggregator platforms. Panchal said the new generation will only buy online, so we have to be online in order to survive.

There was also an internal worry: revenue and expenses were both growing in a straight line, and if this continued, expenses would overtake revenue. Historically, expenses had in fact grown more slowly than revenue. What they feared was the next chapter — that the physical-network route was about to lose the operating leverage it had had in the past.

In November 2019, they wrote down the "omnipresent" framework. A few months later, the pandemic arrived. Jain called in the top 100 leaders and said: after the pandemic the economy will shrink, and we will get smaller too, but we are not organised for that smaller size. Imagine you have US$1 billion of capital, and tell me how you would rebuild your department. He later recalled that everyone knew the answer was to accelerate digitisation and become a consumer-centric company. BFL handled 30% of India's electronics and 15% of its smartphones. Going online could give its own merchants a path that didn't depend entirely on Amazon, while building up data for lending.

They mapped every offline process, deleted the steps that added no value, and then put them into the app. The people designing the interface sat inside the businesses, because ideas grow out of the business. A customer who left halfway through in any channel could pick up where they left off in the same channel or a different one. The marketplace had more than a million products, zero-interest loans and 24-hour delivery. Shopping, bill payments, travel and cashback were there to make people open the app again and again.

Which brings us back to the question in the conference room.

1. The 50.5 million existing customers hold 45% of India's outstanding mortgages, 65% of its credit cards, and US$100 billion in personal loans. BFL captures only about 10% of their wallet. Going deeper requires almost no new capex.
2. They still need to add 6 to 7 million new customers a year. Appliances are replaced slowly, and cross-selling can't make up that growth. Without new geographies, competitors will catch up.
3. Distribution covers only 62% of Indian households; the other 38% have never been touched. The app could become a partner for small-town retailers facing e-commerce. But there is little credit history there, more reliance on franchisees, service will be uneven, and customers unfamiliar with formal repayment may bring higher defaults.

Bajaj says they have a history of reinventing themselves in the face of adversity. This time, the opponent is Amazon.

My Thinking

You can compete with Amazon — as long as you don't try to become a second Amazon

Yes, it can compete. But the layer it competes on is the credit decision and repayment, not the shelf.

In 2020, e-commerce was only 5.8% of Indian retail, and Amazon held 36% of that e-commerce layer. Multiply them and you get about 2% of all retail. BFL stands at the other end: the store, the approval, the collection. It handles 30% of electronics and 15% of smartphones, with 50.5 million customers; Amazon Pay Later had 2 million at the time. A net interest margin of 13.9%, against around 4% for banks. That's the moat — and also the target everyone is aiming at. Non-banks can't get current-account deposits. Amazon can use its e-commerce profits to subsidise buy-now-pay-later. Fighting a price war with a 13.9% margin would mean blowing yourself apart.

So my choice is to take both roads, with the order fixed in stone, and to explicitly not take a third.

First, guard the moment of purchase. An application half-filled in a store must be continuable in the app. Someone already standing at the counter shouldn't end up finishing their next purchase on somebody else's phone.

Then treat the 10% wallet share as profit, not as a substitute for growth. Cross-sellable customers grew only from 24.1 million to 27.4 million in two years, while new customers come in at 6 to 7 million a year. Nobody buys a fridge every quarter. Panchal says using cross-selling to replace new customers is "easier said than done," and I agree.

The 38% of uncovered households have to be reached. But go there carrying the approval rules already proven in the cities, not the enthusiasm of franchisees. In June 2021, about 70% of rural locations were franchises. In 2008 they shut 330 locations and dropped 14,500 dealers to bring net NPAs from 7% down to 0.12%. There's no need to pay that tuition again. Until defaults on franchisee-originated loans line up with those of company-owned branches, their limits should be capped.

Shopping, bill payments and travel should serve only as reasons to open the app. Of people buying appliances online, 52% are after a lower price and 46% after convenience. The case also says that 90% of consumers prefer aggregator platforms over a brand's own app. A price war will be lost. What can win is "zero interest, next-day delivery, a warranty" — and only on the purchase that needs financing. The moment the marketplace becomes another Amazon, BFL is competing for business with its own 114,000 retail partners.

Amazon is buying the next time you open the app. What BFL has to protect is that the door is still there the next time you want to borrow.

I accept the cost of this choice: some towns will be reached a year later. In FY2021, provisions rose from US$519 million the year before to US$816 million, and return on equity fell from 20.2% to 12.8%. In FY2019 it had still been 22.5%. What blew a hole in that year was credit cost, not expenses. Slow is better than paying for this lesson a second time.

My own inclination: if I'm forced into an either-or in class, I'll pick "wallet first, then small towns," not "only dig into existing customers." Halting the 6 million new customers a year is handing the next town to whoever shows up first on people's phones.

The year digitisation really began was 2015, when the data came back to the company

The app is the final chapter. Before it come seven decisions, and next to each one is a pit.

Clean up the books first, then talk about scale. Expanding with net NPAs at 7% means diluting bad debt with growth. They did the reverse, cutting urban consumer durables stores from 19,000 to 2,000, then rebuilding bit by bit to 26,900. That cut was strategy, not a clean-up.

Change the game to same-day delight, not lower prices. Three to five days brought down to 3 hours. The pit next to speed is having no rules — bad debt was at one point three times the industry level. The rules were later put into software; change one, and it takes effect across every location.

Only do what's big enough to be worth doing, and shut the rest. Fifty products sounds like Olay. The difference is that things get shut down: infrastructure lending exited, life insurance sales that couldn't take off were stopped, new businesses double, mature ones grow at least 20%. Pamnani's "don't develop hobbies" is product strategy, not a description of personality.

In 2015, customer data was defined as belonging to the company, not as a department's crown jewels. An order alone wasn't enough. Business units using others' customers had to pay a transfer fee, and the source — consumer durables and mobile financing, which brought in 90% of the customers — got 35% of its profit from it. Sharing became revenue, and people were willing to hand things over.

The arms and legs can be outsourced; the brain stays in-house. Approval rules and process design don't go into a vendor's black box — nor do they build every system themselves.

Approval, risk and collections are split into three jobs, with the latter two jointly bearing losses. Retailers can see their own approval rates. When the rate drops, they bring in better customers on their own. Control doesn't have to sit at headquarters.

Only last comes 2019. During the pandemic, those 100 people were asked to rethink their departments for "a smaller economy and US$1 billion of capital." The head of technology admitted they had always been B2B and had hardly built a consumer-facing layer at all. The processes were mapped first, the steps that added no value deleted, and only then did it all go into the app. The people building the interface sat inside the business.

What they avoided weren't technology pits, but incentive pits. Handing over data earned you money, bad debt couldn't be hidden, and businesses that didn't work couldn't survive. There were also three kinds of grown-ups in the room: an owner, a person who had spent a lifetime lending, and an operator. And a scar. A company that has seen 15% bad debt doesn't put risk in an appendix.

My own reading is that many companies understand "digitisation" as a launch date. If BFL could only remember one date, it should be the year the data came back to the company, not the year customers could download the app. Without that earlier step, the app just moves the silos online.

This is a company that deeply understands the moment of purchase — but not yet a customer-centric company

At the moment of purchase, they are serious. Parents want the TV on the wall the same day. People who aren't in a hurry for cash want good service, so salaried loans can keep the algorithm in the back office while the front desk still approves. One rupee brings people through the door, and it's the customers themselves who push up the basket size.

The organisation, though, runs on a different logic. Each product has a mini-CEO. Cross-selling is pushed by a separate, heavily incentivised team, with risk pre-approved. An existing customer can't walk up to this team on their own and ask for another loan. The company decides when a customer is allowed to buy more.

After the pandemic, Jain said the answer was to accelerate digitisation and "become" a consumer-centric company. If you're becoming one, you weren't one when you set out. The head of technology said it more bluntly: they did B2B for many years and never seriously built the consumer-facing layer.

That line, "don't pay twice for the same customer," is a sentence about acquisition cost. Here, the customer is an asset. And the asset is very well managed.

My own view is that customer-centricity isn't about making the experience smooth. Smooth they've already achieved. Customer-centricity is when the customer can walk into the next product on their own, rather than waiting for a team they can't reach to push it to them. Amazon describes itself as "most customer-centric." BFL describes its ambition as "omnipresent" and manages itself by ROE. Both statements are true. They are not the same statement.

Loyalty stops at zero interest and that card

The EMI card does bring real repeat purchases. For people who can't get a bank card, it's the only door. People come back to a door that doesn't exist anywhere else. That's loyalty through access.

Zero interest is another matter. People who can afford to pay in full take it anyway. While the terms are there, the people are there. When another counter writes the same zero-interest offer, people no longer need to be.

The wallet tells the rest of the story. These customers hold 45% of India's mortgages, 65% of its credit cards and US$100 billion in personal loans, and BFL accounts for only about 10%. Home finance is just 12% of its own loan book. That 45% of mortgages wasn't lent by BFL. Where customers truly have a choice, nine rupees out of ten go elsewhere.

Then look at the pace. Six to seven million new customers a year, yet the cross-sell pool grew by only about 3.3 million in two years. About 30% of new customers are first-time credit users. This is a superb customer acquisition machine. A business compounding on loyalty wouldn't need this many strangers to keep growing.

People repeat something that's convenient and unavailable elsewhere. Once the convenience is available elsewhere, the repeating stops.

The TV goes up on the wall the same day, and the delight is real. After the delight settles, it will be a long time before this company is needed again. Bill payments and travel are there to fill that gap. Fill it clumsily and it becomes a hobby; don't fill it, and the next time the customer wants to buy something, they open somebody else's app.

Get one ledger in order close to home before talking about being everywhere

If I could take only one thing away from this case, it would be the sequence.

Close to home: first consolidate each customer into a single ledger, and make sure the people who hand over that ledger get credit for it. BFL had the warehouse and the transfer fees first, and the app afterwards. Do it in reverse, and digitisation just asks the same questions again, faster.

In the middle: map the process first, delete the steps that don't change any decision, then write the system. Split speed and standards into two jobs, and make both accountable for losses. Before starting anything new, write down the criteria for shutting it down, and give it a date. A rule without a date never happens.

Further out: being everywhere isn't an app with more goods hanging on it. It's still being there the next time someone truly needs you. For a consumer durables finance company, the next time might be years away. The same goes for any business with a long natural cycle: if frequency isn't designed, the relationship will go dormant on its own.

When competing with a platform, hold on to the judgement it can't yet make. BFL's judgement is: if this person borrows, will they pay it back? The shelf, logistics, whose price is lower — that's a battlefield someone else already holds.

My own inclination is to treat "customer numbers are rising" as a figure that needs correcting by wallet share. The 50.5 million is real. So is the 10%. A company that reports only the first number will think it already owns these customers.

Takeaways for Arcus42

Reading Bajaj, what came to mind wasn't lending in India, but something more ordinary: we too are very good at bringing someone through the door, and then assuming they'll stay.

First, the interface is the last step; the file is the first. If the facts about a client are still scattered across different people, different spreadsheets, different conversations, building a new entry point just moves the scattering online. First consolidate it into one file, assign an owner, and make it possible for a second person to explain this family clearly without gathering everything again. If we can't do that, the "omnipresence" that follows has nothing at its core.

Second, whoever hands over a client must get credit for it. BFL didn't tear down departmental walls with speeches. Departments using others' customer data had to pay, so the source was willing to keep acquiring customers. Inside a company, whoever builds the trust should still be visible when the next piece of work is done by someone else. Otherwise everyone will keep a ledger that belongs only to them.

Third, move fast, but don't let standards loosen along with the pace. Three-hour approvals didn't slide back to 15% bad debt because the rules were in a different group's hands, and losses were shared. The relationship can move quickly. The threshold for judgement doesn't drop because the other side is in a hurry, or because we want to close. The threshold is best made public, so people further up the line can screen themselves, rather than being turned down at the very end.

Fourth, the time between meetings has to be designed; you can't count on the transaction itself to keep people. Zero interest brings people through the door. Behind the door is a long gap. A family office's natural cycle is even longer: an inheritance, a liquidity event, the next generation coming in. If we only show up at those moments, we are that zero-interest loan: there when the terms are good, gone when the terms aren't the only thing that matters. Frequency should come from the work itself — reviews, governance, the next generation in the room — not from bolting on more features unrelated to the core business.

Fifth, don't copy the platform's shelf in order to compete with the platform. The moment BFL turns its marketplace into another Amazon, it loses both its margin and its retail partners. Our reason for people to stay should be that the judgement, the records and the trust are here, that the product serves a family's own allocation — not that we happen to offer every item on somebody else's menu as well. A longer menu is not presence.

Final Thoughts

The pre-class me most agrees with this line: "What you win is the moment of the sale; what lies empty is the space between two sales."

After class, I'll look at which of these got overturned, and by what.

---

These are pre-class reading notes, compiled from Harvard case materials, and do not constitute investment advice on Bajaj Finance or any financial product.