Daybreak - How AI turned banks’ risk data into advertising

Episode Date: November 9, 2025

Across India, lenders like HDFC and Kotak are repurposing the same algorithms that once judged credit risk to run hyper-personalised marketing campaigns. These systems now predict who’s ready for a credit card, or insurance plan by studying every transaction, payment, and habit — turning customer data into personalised sales pitches.The RBI has drawn clear lines for how AI can be used in lending. But when it comes to marketing, the rules are still fuzzy. That leaves space for India’s biggest banks to experiment—and for AI to quietly reshape how finance sells itself.Tune in.Daybreak is produced from the newsroom of The Ken, India’s first subscriber-only business news platform. Subscribe for more exclusive, deeply-reported, and analytical business stories.

Hi, this is Rohan Dharma Kumar. If you've heard any of the Ken's podcasts, you've probably heard me, my interruptions, my analogies, and my contrarian takes on most topics. And you might rightly be wondering why am I interrupting this episode too? It's for a special announcement. For the last few months, I and Sita Raman Ganeshan, my colleague and the Ken's deputy editor, have been working on an ambitious new podcast. It's called Intermission.

We want to tell the secret sauce stories of India's greatest companies. Stories of how they were born, how they fought to survive, how they build their organizations and culture, how they manage to innovate and thrive over decades, and most importantly, how they're poised today. To do that, Sita and I have been reading books, poring over reports, going through financial statements, digging up archives, and talking to dozens of people. And if that wasn't enough, we also decided to throw in video into the mix. Yes, you heard that right. Intermission has also had to find its footing in the world of multi-camera shoots in professional studios, laborious editing and extensive post-production. Sita and I are still reeling from the intensity of our first studio recording.

Intermission launches on March 23rd. To get an alert as soon as we release our first episode, please follow Intermission on Spotify and Apple Podcasts or subscribe to the Ken's YouTube channel. You can find all of the links at the ken.com slash I am. With that, back to your episode. When Shilpa got her first business loan from Kotak Mahindra Bank in 2024, she thought her relationship with a bank would quiet town, limited to just the regular EMI's. Instead, it got chatty.

Five years ago, every bank had turned her away. She had no civil score, no formal collateral, and no guarantee that she would return the money. But this time, Kotak had approved the loan in days. Its artificial intelligence tool sifted through her payment history, store location and transaction patterns and decided she was worth a few lakhs. Soon, Shilpa opened her tailoring shop in East Bangalore.

But ever since, her phone hasn't stopped buzzing. Every few days, she gets a new message, secure your future, buy loan insurance, or turn your business into a brand. Sometimes, it's a video about SIPs for new entrepreneurs. Other times, nudges that encourage her to start investing. The 34-year-old told the Ken reporter Muromai that before banks only pushed home or car loans.

Now it's about business insurance, which actually makes sense for her. You see, it wasn't anything that Shilpa did that changed. It was her bank. The same data that once judged whether she deserved credit now decides what else she might want to buy. And that shift isn't accidental. It's happening all across India's financial sector.

Banks like HDFC, Kotaq and microfinance firms such as Chaitanya India, FinCredit, are repurposing their AI engines. These engines, which are built. on years of transaction and behavioral data are now pushing hyper-personalized campaigns. Take HDFC Bank's habitual AI, for instance. It was an experiment that nudged inactive users to start transacting again, and conversions jump by 20%. There's also KOTU's new marketing OS,

which connects data across savings, loans and investments to spot future cardholders. And at Chaitanya, AI micro-segmentation campaigns helped find small town retail ready for another loan. For context in marketing, micro segmentation means splitting customers into very small, specific groups so brands can send each one more personal targeted messages.

Of course, for banks, the economics are very tempting. As one ad tech founder put it, while credit risk AI saves time, marketing AI makes money. Even consulting firm McKinsey says companies using AI for marketing see 20 to 30% higher returns

than old school. methods. But there's a catch. You see, Reserve Bank of India strictly regulates AI in lending. But in marketing, not so much. Consent boxes stay conveniently ticked, data slips easily between departments and the nudges never really stop. And the result, AI is now turning risk into a product. Welcome to Daybreak, a business podcast from the Ken. I'm your host Rachel Vergis and every day of the week, My co-host, Nika Sharma and I, will bring you one new story that is worth understanding and worth your time. Today is Monday, the 10th of November.

HDFC Bank thought it knew all 92 million of its customers. But then, a senior executive from the bank told Miramai that they discovered something odd. What they found was that while salaried customers in smaller towns, you know, places like Nagpur and Mysur, were steady savers. They rarely touch the bank's investment products. Fortunately for them, the same senior executive found the fix to be surprisingly simple. Basically, to keep the customers from drifting away, HDFC partnered with a tech firm called Three Lock.

Together, they built an AI tool that nudged inactive app users to explore new features. The data science team also added look-alike models, which are basically digital twins of those users and launched several AI-triggered campaigns. Think push notifications about saving goals, credit card. discounts on bill payments or quick videos showing how small SIPs could fund a holiday abroad. You see, that's the real shift. Data that once kept bankers awake at night, like risk reports, payment histories, spreadsheets full of numbers, is now helping them sell products.

The same information that told a credit officer, if someone was safe for a loan, now tells marketing engines how to make them buy one. And this new AI-driven MARTEC or marketing tech isn't just saving banks money. It's opening up brand new markets. You see, earlier traditional marketing used to divide customers based on broad demographics. Age, income, city, job title, messages were standardized and not personalized. AI is now changing that. The executive told us that they had their vendors study people by behavior.

They found micro segments. Basically profiling people based on frequent travels, people who spend very little at a time, or people who pay rent but have no fixed salary. What once took weeks of planning now happens in minutes. Now, while HDFC refined how it spoke to customers, Kotak Mahindra focused on who it spoke to. The bank built what it calls a semi-autonomous marketing engine, which Kedarswamy Ravay, Kotak's head of marketing,

said was their most exciting AI use case. Another executive who works on Kotak's AI-led marketing engine told us that earlier they ran 10 different campaigns on 10 different data sets. And then, once they unified all customer data, savings, lending, investments, the patterns were obvious. AI could spot not just who had a good credit score, but who behaved like a future cardholder. The result? A wave of first-time credit card users from places that rarely showed up on old dashboards, small town business owners in Nagpur, Coimbutur and beyond. And now, microfinance institutions are also getting in on it.

At Chaitanya India FinCredit, a huge data cleanup originally meant for lending decisions revealed new marketing opportunities. So, instead of blasting generic campaigns that work for everyone, their AI focuses on moment-based personalization, basically tracking customer habits to nudge them towards small loans or investments. And even though it's still early days, a data analyst there said that Chaitanya saw about a 15% uptick in conversions for insurance and investment products. Now, in theory, AI makes decisions fairer.

It can weigh many factors like Shilpa's location and payment history and judge context better than a simple rulebook. But there's a risk as well. It can inherit human bias. For example, if a bank's old data avoided certain towns or lent less to women or favored salaried workers, the AI learns that pattern and repeats it. It doesn't ask why the pattern exists. It just learns that it worked once and then keeps doing it. Stay tuned. N. Avinash runs a small retail store in Mandya, Karnataka and uses Chaitanya for loans.

But lately, his banking app has been trying to sell him dreams. He told us that while it's not like they're selling new things all the time, the videos are different now. The first one he remembers was cute, an animation of a man sewing coins that grew into plants with money. But the next one hit a little closer to home. He said he saw a video of a guy buying his first light truck. And it stuck with him because he'd been thinking of buying one himself. That's exactly how today's marketing AI works. Precise and very persistent. But that precision also raises a bigger question. When is a nudge educational and when does it become a shove? And who judges the difference?

Now, the Reserve Bank of India has started asking similar questions. It recently introduced something called Free AI, which is short for the framework for responsible and ethical enablement of artificial intelligence. It has 26 recommendations across six pillars, all aimed at preventing harm and misuse of data in credit risk assessment. So, when banks use use, use AI to decide who gets a loan, they're closely monitored. But when they use the same tech for marketing, it's not quite as well monitored.

You see, AI in marketing is currently governed by the IT Act and soon the upcoming Digital Personal Data Protection Act or DPDP. Under that law, marketers will need explicit consent from users to process or analyze their data. It will also make companies responsible for any leaks. But personal data can mean almost anything. Basically, anything that identifies you as an individual counts as personal data. So, a marketing vendor might not know your name or address, but they'll know how much you earn, how often you travel, or whether you pay rent.

And that's more than enough for them to sell you like travel insurance. The thing is, no one really knows how much of the BSFI or banking financial services and insurance sector's marketing spend goes into AI-driven campaigns. But a 2025 Boston Consulting Group report estimates that Indian CXOs plan to raise investments in AI marketing to 60% within the next few years. And it's easy to see why. BG Mahesh, CEO of Industry Body Sahamati, said that AI in Martec is actually low-hanging fruit. There's zero regulatory friction, faster ROI and tons of customer data to work with.

On the other hand, he said, credit models need expectations. plainability and compliance. Marketing doesn't. It's easier to optimize who clicks on a loan ad than to reinvent how the loan itself gets approved. So far, there's no clear evidence of mis-selling. But the line between helpful and manipulative is thin. In fact, Mahesh warned that AI marketing risk crossing that line.

He added that the next phase of AI and BFSI must shift from predicting behavior to enabling better choices. The challenge now is to build. ethical AI frameworks that empower, not exploit users. Still, the current scenario is favourable to at least one group, at the companies. More on this in the next segment. The HDFC executive told Miran Mai that the future of advertising is personalisation. Earlier, they'd have to wait weeks for a campaign.

Now, the system tells him which line worked best within hours. In fact, he claimed that the bank's marketing department is increasingly relying on ad tech platforms rather than creative agencies. That's because they analyze ad performance data and customer transaction patterns and then tell banks which of the creatives or keywords led to a particular purchase. Basically, they tell banks what works and where. Rajiv Dingra, the founder and CEO of AdTech firm rebid, said that the system only works better when you put agentic AI on top of it.

It surfaces insights that no human can spot in mere minutes. Dingra believes that in marketing there's no negative impact. He added that an error rate of 10 to 15% is normal and even 30% is acceptable. Basically, the most you could lose by showing an unnecessary ad is just some of your marketing budget. But when it works, it works really well.

For instance, rebid system once spotted a spike in small loan leads near Chennai. When the bank checked, it found a new factory had opened there, driving up small loan demand in the area. The bank eventually cashed in on that opportunity. You see, in India, that kind of precision is very useful. Ruturaj Biswas, co-founder of AdTech company HyperGrow, said that even a few pieces of data, say how often someone books flights,

can help At-Tech firms build a credible customer profile. The appeal, of course, is obvious. As Dingra said, India has one of the world's largest working-age populations. And of course, every BFSI player is chasing this growing market. But they can only win if their targeting is right the very first time. And so, the targeting continues. And like any other profile, somewhere inside these dashboards and data models sits Shilpa, the tailor from Bangalore stitched together by overlapping data points.

The same systems that helped her grow her business are now teaching themselves how to sell her more. She hasn't bought loan insurance yet, but she has started an SIP. Tomorrow, the model will tell the bank what else she's ready for. A car, a home, a second shop even. And unless someone decides where ready ends and risk begins, the AI won't stop recommending. A machine, after all, doesn't question intent. It only optimises it.

Daybreak is produced from the newsroom of the Ken, India's first subscriber-focused business news platform. What you're listening to is just a small sample of our subscriber-only offerings. A full subscription offers daily long-form feature stories, newsletters, and a whole bunch of premium podcasts. To subscribe, head to the ken.com and click on the red subscribe button on the top of the Ken website. Today's episode was hosted and produced by my colleague Rachel Vargis and edited by Rajiv Sien.