In 2016, a friend of mine who had spent years in traditional banking left his job at a private bank to join one of India’s first digital lending startups. He was excited in a way I had not seen from him before. He kept saying the same thing in different ways. We are going to fix this. No more branch queues, no more agents pushing products nobody needs, no more decisions made on a relationship manager’s mood that day. Just data, speed, and fairness, built into the product itself.
Four years later, he left that company too. Not because the technology failed. The technology worked exactly as designed. He left because the growth targets had quietly become the only thing anyone in leadership actually measured, and the fairness he had signed up for had become a slide in the pitch deck rather than a line in the product.
He is not a unique case. He is, in many ways, the story of an entire generation of Indian fintech.
It is worth remembering what digital lending promised, because the promise was genuine, and largely still is technically true. Branches meant geography decided who got credit. Digital lending removed the branch. Paperwork meant only the already-documented, already-banked could access formal credit. Digital lending, especially once layered onto Aadhaar and the Account Aggregator framework, meant a much larger set of Indians could be seen by a lender at all. Manual underwriting meant slow, inconsistent, sometimes biased decisions. Automated underwriting meant faster, theoretically more consistent decisions at a scale no branch network could match.
Every one of these was a real improvement over what came before. This article is not an argument that digital lending failed to deliver access. It delivered access at a scale traditional banking never could have.
The failure is narrower, and more specific. Access was delivered. Fairness was not required to travel with it.
Here is the pattern that repeats across nearly every fintech lending company that started with a genuine mission and ended up looking uncomfortably similar to the industry it set out to replace.
At the beginning, the founding team is close to the product and close to the user. Decisions get made by people who still remember why the company exists. Growth is a proof point, evidence the model works, not yet the only thing that matters.
Then capital arrives, and with it, a specific kind of pressure that is rarely discussed honestly. Growth stops being a proof point and becomes the product. A Series B, a Series C, is not evaluated on how well the company served its existing users. It is evaluated on how fast the loan book grew since the last round. Nobody explicitly decides to deprioritise the user. The org chart simply starts rewarding the people who hit disbursal numbers, and the people who ask harder questions about whether a specific loan was right for a specific person get quietly sidelined, because their instinct slows down the number that matters to the board.
By the time a company is at scale, the incentive structure that shaped it, get more people approved, disburse more, at almost any acceptable default rate, has become the water everyone in the building swims in. Nobody working there today necessarily thinks of themselves as running an extraction machine. They think of themselves as running a growth business. The mechanism that got them there simply stopped asking the one question that matters, and the culture has forgotten it was ever supposed to be asked.
This is not a story about bad people. My friend who joined that lending startup in 2016 was not a bad person, and neither were the people he worked with. It is a story about what happens when a system’s only feedback loop is a number that has nothing to do with the person on the other end of the loan.
There is an uncomfortable earlier chapter in Indian financial services that digital lending has, in many ways, quietly repeated.
Life insurance mis-selling in India through the 2000s and early 2010s followed almost exactly this arc. ULIPs and endowment policies with enormous upfront commissions, sold by agents whose income depended entirely on the size and frequency of the sale, not on whether the policy served the buyer. Regulators eventually stepped in, capped commissions, mandated disclosures, restructured the entire distribution model. It took years, and it took a great deal of damage to consumer trust in the meantime.
Digital lending’s DSA and agent-driven distribution, the subject of RBI’s most recent regulatory directions, is close to a rerun of the same structural problem in a newer wrapper. Commission slabs and payout accelerators that reward volume, agents whose only real metric is disbursal, a channel that scaled faster than any single institution’s ability to personally watch every interaction. RBI’s own recent directions describe this almost exactly, noting that accountability naturally thins the further a sale gets from the lender’s own building, no matter how well run the institution is.
The pattern is the same. Only the product changed, from an insurance policy to a loan.
Nearly every fintech lending company in India has, at some point, used language about being user-first, transparent, and fair in its marketing. Very few have built the actual structural mechanism that would make those words true under pressure.
There is a simple test for whether a company’s fairness commitment is structural or rhetorical. Ask what happens to an employee’s compensation, promotion, or standing if they say no to a sale that would have hit a growth target but was wrong for the customer. In a structurally fair company, saying no is rewarded, or at minimum, never punished. In most fintech lending companies, saying no quietly costs that employee something, a missed target, a harder quarter, a slower path to the next level, even if nobody ever says this explicitly out loud.
Marketing language does not survive that test. Only compensation structures, reporting lines, and what actually gets measured survive that test.
This is the deepest reason the fintech promise broke down. Not bad intentions at the founding. Not a lack of genuine belief in the mission. A failure to build the org design that would have protected the mission once growth pressure arrived, which it always does, for every company that raises capital.
RBI’s Amendment Directions on advertising, marketing, and sale of financial products, issued in June 2026 and effective January 2027, read, in places, like a regulator finally naming this exact failure mode in writing.
The directions draw a formal distinction between eligibility and suitability, whether a borrower qualifies for a loan is a different question from whether that specific loan is right for that specific person, and require lenders to actually assess the second, not just the first. They bring commission structures, the slabs and accelerators that reward volume over fit, inside the compliance perimeter, meaning a lender’s payout grid is no longer a private business decision but a supervised surface. And most pointedly, they state that a sale can be mis-selling even with the customer’s full, explicit consent, if the product itself was unsuitable for that customer’s profile.
That last point is the regulator closing the exact loophole this article has been describing. Consent has been the industry’s long-standing defence, the customer signed, so the company is covered. RBI has now said, in effect, that consent alone was never a sufficient defence if the underlying incentive structure was built to extract rather than serve.
It has taken the regulator years to arrive at a rule that says, in formal language, what should have been obvious from the beginning. A company’s structure should make it costly to sell someone something wrong for them, not just legally survivable.
This is not an argument that every fintech lending company in India operated in bad faith, and it is not an argument that digital lending should not have happened. It should have, and it delivered genuine access that traditional banking structurally could not.
It is an argument that access without a structural commitment to fairness eventually curdles into exactly the extraction it was meant to replace, and that the only real defence against this is not better marketing language, but compensation and incentive structures built, from the beginning, to survive the pressure that growth capital always brings.
The next article in this series turns from diagnosis to construction. If commission-driven distribution, growth-at-any-cost incentives, and consent-as-a-defence are the structural failures that got us here, what does a lending model built the other way around actually look like in practice, not as a mission statement, but as an org chart, a compensation structure, and a product.
Previous in the series, The User Nobody Asks, What Borrowers Actually Experience
Next in the series, What User-First Lending Actually Looks Like
In 2016, a friend of mine who had spent years in traditional banking left his job at a private bank to join one of India’s first digital lending startups. He was excited in a way I had not seen from him before. He kept saying the same thing in different ways. We are going to fix this. No more branch queues, no more agents pushing products nobody needs, no more decisions made on a relationship manager’s mood that day. Just data, speed, and fairness, built into the product itself.
Four years later, he left that company too. Not because the technology failed. The technology worked exactly as designed. He left because the growth targets had quietly become the only thing anyone in leadership actually measured, and the fairness he had signed up for had become a slide in the pitch deck rather than a line in the product.
He is not a unique case. He is, in many ways, the story of an entire generation of Indian fintech.
It is worth remembering what digital lending promised, because the promise was genuine, and largely still is technically true. Branches meant geography decided who got credit. Digital lending removed the branch. Paperwork meant only the already-documented, already-banked could access formal credit. Digital lending, especially once layered onto Aadhaar and the Account Aggregator framework, meant a much larger set of Indians could be seen by a lender at all. Manual underwriting meant slow, inconsistent, sometimes biased decisions. Automated underwriting meant faster, theoretically more consistent decisions at a scale no branch network could match.
Every one of these was a real improvement over what came before. This article is not an argument that digital lending failed to deliver access. It delivered access at a scale traditional banking never could have.
The failure is narrower, and more specific. Access was delivered. Fairness was not required to travel with it.
Here is the pattern that repeats across nearly every fintech lending company that started with a genuine mission and ended up looking uncomfortably similar to the industry it set out to replace.
At the beginning, the founding team is close to the product and close to the user. Decisions get made by people who still remember why the company exists. Growth is a proof point, evidence the model works, not yet the only thing that matters.
Then capital arrives, and with it, a specific kind of pressure that is rarely discussed honestly. Growth stops being a proof point and becomes the product. A Series B, a Series C, is not evaluated on how well the company served its existing users. It is evaluated on how fast the loan book grew since the last round. Nobody explicitly decides to deprioritise the user. The org chart simply starts rewarding the people who hit disbursal numbers, and the people who ask harder questions about whether a specific loan was right for a specific person get quietly sidelined, because their instinct slows down the number that matters to the board.
By the time a company is at scale, the incentive structure that shaped it, get more people approved, disburse more, at almost any acceptable default rate, has become the water everyone in the building swims in. Nobody working there today necessarily thinks of themselves as running an extraction machine. They think of themselves as running a growth business. The mechanism that got them there simply stopped asking the one question that matters, and the culture has forgotten it was ever supposed to be asked.
This is not a story about bad people. My friend who joined that lending startup in 2016 was not a bad person, and neither were the people he worked with. It is a story about what happens when a system’s only feedback loop is a number that has nothing to do with the person on the other end of the loan.
There is an uncomfortable earlier chapter in Indian financial services that digital lending has, in many ways, quietly repeated.
Life insurance mis-selling in India through the 2000s and early 2010s followed almost exactly this arc. ULIPs and endowment policies with enormous upfront commissions, sold by agents whose income depended entirely on the size and frequency of the sale, not on whether the policy served the buyer. Regulators eventually stepped in, capped commissions, mandated disclosures, restructured the entire distribution model. It took years, and it took a great deal of damage to consumer trust in the meantime.
Digital lending’s DSA and agent-driven distribution, the subject of RBI’s most recent regulatory directions, is close to a rerun of the same structural problem in a newer wrapper. Commission slabs and payout accelerators that reward volume, agents whose only real metric is disbursal, a channel that scaled faster than any single institution’s ability to personally watch every interaction. RBI’s own recent directions describe this almost exactly, noting that accountability naturally thins the further a sale gets from the lender’s own building, no matter how well run the institution is.
The pattern is the same. Only the product changed, from an insurance policy to a loan.
Nearly every fintech lending company in India has, at some point, used language about being user-first, transparent, and fair in its marketing. Very few have built the actual structural mechanism that would make those words true under pressure.
There is a simple test for whether a company’s fairness commitment is structural or rhetorical. Ask what happens to an employee’s compensation, promotion, or standing if they say no to a sale that would have hit a growth target but was wrong for the customer. In a structurally fair company, saying no is rewarded, or at minimum, never punished. In most fintech lending companies, saying no quietly costs that employee something, a missed target, a harder quarter, a slower path to the next level, even if nobody ever says this explicitly out loud.
Marketing language does not survive that test. Only compensation structures, reporting lines, and what actually gets measured survive that test.
This is the deepest reason the fintech promise broke down. Not bad intentions at the founding. Not a lack of genuine belief in the mission. A failure to build the org design that would have protected the mission once growth pressure arrived, which it always does, for every company that raises capital.
RBI’s Amendment Directions on advertising, marketing, and sale of financial products, issued in June 2026 and effective January 2027, read, in places, like a regulator finally naming this exact failure mode in writing.
The directions draw a formal distinction between eligibility and suitability, whether a borrower qualifies for a loan is a different question from whether that specific loan is right for that specific person, and require lenders to actually assess the second, not just the first. They bring commission structures, the slabs and accelerators that reward volume over fit, inside the compliance perimeter, meaning a lender’s payout grid is no longer a private business decision but a supervised surface. And most pointedly, they state that a sale can be mis-selling even with the customer’s full, explicit consent, if the product itself was unsuitable for that customer’s profile.
That last point is the regulator closing the exact loophole this article has been describing. Consent has been the industry’s long-standing defence, the customer signed, so the company is covered. RBI has now said, in effect, that consent alone was never a sufficient defence if the underlying incentive structure was built to extract rather than serve.
It has taken the regulator years to arrive at a rule that says, in formal language, what should have been obvious from the beginning. A company’s structure should make it costly to sell someone something wrong for them, not just legally survivable.
This is not an argument that every fintech lending company in India operated in bad faith, and it is not an argument that digital lending should not have happened. It should have, and it delivered genuine access that traditional banking structurally could not.
It is an argument that access without a structural commitment to fairness eventually curdles into exactly the extraction it was meant to replace, and that the only real defence against this is not better marketing language, but compensation and incentive structures built, from the beginning, to survive the pressure that growth capital always brings.
The next article in this series turns from diagnosis to construction. If commission-driven distribution, growth-at-any-cost incentives, and consent-as-a-defence are the structural failures that got us here, what does a lending model built the other way around actually look like in practice, not as a mission statement, but as an org chart, a compensation structure, and a product.
Previous in the series, The User Nobody Asks, What Borrowers Actually Experience
Next in the series, What User-First Lending Actually Looks Like
