Five habits behind a pledged-gold product

Most of our borrowers complete their first loan with a branch officer, and many prefer to. I design the tablet as a shared screen. The appraiser weighs the gold and tests its purity in front of the customer, and the tablet shows weight, purity, the value our loan-to-value limit allows and the amount offered. The key fact statement appears in the customer's chosen language, and the officer reads its summary aloud before signature. I measure counter time, but I also measure whether a customer can repeat the interest rate and the repayment date when we call a week later. When the two measures conflict, I hold the explanation. Credit sets the limits and appraisal standards; my team decides how those limits are shown and what the officer must say.

Our customers read Gujarati, Hindi and Marathi far more comfortably than English, and some prefer to listen rather than read. Four language editors work inside the product team. They do not translate finished English text; they write each notice with the product manager and test it by asking borrowers to explain it back in their own words. A notice passes when most borrowers in a small session can say what is owed, by when and what happens next. We keep a glossary so that the words for interest, renewal and auction do not change between the app, the SMS and the branch script. Compliance reviews every regulatory notice in each language, not only the English version that would otherwise become the authority. Recorded voice versions go through the same test.

Gold-loan fraud takes several forms: spurious gold at the counter, staff collusion, impersonation of a borrower and scams in which a borrower is tricked into acting against herself. Appraisal and branch controls belong to risk and audit colleagues. My responsibility is the fraud that a product makes possible. I ask what a fluent scammer could persuade a frightened customer to do on our app, then add friction where that persuasion would work. Holds after a change of device, call-backs from the home branch and confirmation screens that state the amount in words are cheaper than reimbursing losses. They also cost genuine customers time, so I report both effects. I review complaint registers and branch observations weekly, because the first signal often arrives at a counter rather than in a dashboard.

When a gold loan is not repaid, the lender may eventually auction the pledged gold. The RBI expects proper notice to the borrower and a transparent auction process, and our legal and collections teams own that procedure. My team owns how the borrower experiences the months before it. We send reminders in her language, give branch staff a call script that explains the options without pressure, and offer interest payment and renewal through UPI or a part-release that returns some jewellery. These choices lengthen resolution and add work at branches. I accept that cost, but I do not let the journey delay the regulatory process indefinitely. Each extension has a written end date, agreed with collections and reported to the risk committee. Borrowers who ask for more time are never told that a reminder will stop an auction.

A product roadmap at a lender tends to grow, because each new feature has an advocate and each retired one has users. I review every item twice a year against three questions: does it help a borrower understand or control her loan, does it create a fraud or conduct risk we cannot supervise, and would branch staff miss it? In 2025 we dropped twelve of 31 planned items, including a gold-price alert that encouraged top-ups. Stopping work frees designers and language editors for the notices and assisted journeys that matter more. It also disappoints colleagues who championed the dropped items, so I explain each decision in writing and record what evidence would bring the idea back. The dropped gold-price alert, for example, would return only if call-back data showed no link to scam complaints.