15.5 Fatal mistakes
1. Chasing volume before credit muscle is built
Section titled “1. Chasing volume before credit muscle is built”The single most common failure mode for new lenders. Sales team is hired ahead of underwriting; loans are pushed through; first vintages default; partner / co-lender confidence collapses; capital is lost.
Refuse: do not hire sales / DSA ahead of underwriting + collections capacity. Volume targets in the first 12 months are deliberately constrained by credit team’s bandwidth.
2. Compliance shortcuts under deadline pressure
Section titled “2. Compliance shortcuts under deadline pressure”KFS slips, consent not captured, NACH not active before disbursement, recovery agent not intimated, bureau submission late. Each is small in isolation; collectively they invite RBI action or borrower-class-action grievance escalation.
Refuse: compliance is a gate, not a checkbox. No production deploy that breaks any compliance gate, no matter the commercial cost.
3. Multi-partner / multi-product at MVP
Section titled “3. Multi-partner / multi-product at MVP”Trying to launch with two co-lending partners, three products, multiple channels at once. Result: nothing is stable; everyone fights fires; no clear performance signal.
Refuse: one product, one channel mix, one (or zero) co-lending partner at MVP. Add complexity only after stability.
4. Aggressive recovery practices
Section titled “4. Aggressive recovery practices”Field agents harassing borrowers; calls outside permitted hours; threats; misrepresentation. One viral incident ends the brand and can revoke licence.
Refuse: zero tolerance. Recordings sampled; complaints actioned; agents removed on first substantiated breach.
5. Ignoring concentration
Section titled “5. Ignoring concentration”Single borrower at 25% of Tier-1, single anchor at > 30% of book, single DSA at 60% of originations, single partner at > 70% of co-lent book. Any of these is one bad week away from collapse.
Refuse: caps enforced in the engine; quarterly review at the board.
6. Under-investing in observability and audit
Section titled “6. Under-investing in observability and audit”System runs; ops are firefighting; no one can answer “why was this loan approved” three months later because the trace was lost. First audit reveals gaps; remediation is expensive; trust is damaged.
Refuse: audit trail is first-class. Audit log integrity validated daily. Decision trace stored. Evidence package retrievable on demand.
7. Stretching DLG / co-lending structures
Section titled “7. Stretching DLG / co-lending structures”Operating at the 5% DLG cap; designing CLM-2 arrangements that economically look like loan sales; using LSP arrangements where the LSP is the real lender. Regulator looks at substance; re-characterisation hurts.
Refuse: operate well within the regulatory letter. Document substance matches form.
8. Hiring ahead of need
Section titled “8. Hiring ahead of need”The temptation to hire 5 more engineers / analysts / RMs because “we’ll need them soon”. Burns cash; creates idle capacity that finds work that shouldn’t be done.
Refuse: hire when the queue + the data justify. Sub-linear staff growth vs book is the marker.
9. Discounting the importance of CA channel
Section titled “9. Discounting the importance of CA channel”CA / Tally / accounting distribution is the structural advantage for this wedge. Treating it as one channel among many — same investment as DSA — wastes the moat.
Refuse: CA channel gets first-class product investment — dedicated portal, joint workflows, fee economics aligned to CA’s economics, accounting-data integration.
10. Productising the platform before the book is proven
Section titled “10. Productising the platform before the book is proven”Trying to sell the platform as SaaS in year 1 because “every NBFC needs this”. External customers expect production-grade; team doesn’t have it; both the SaaS customer and the own book suffer.
Refuse: prove on own + co-lent book to at least ₹100 Cr and 12 months of stable operations before SaaS productisation.
11. Vendor dependence without redundancy
Section titled “11. Vendor dependence without redundancy”Single BSA vendor, single KYC vendor, single AA, single NACH rail. Vendor outage = ops down for hours.
Refuse: every critical primitive has at least a primary and a secondary vendor. Routing rules tested in chaos drills.
12. Skipping the boring infrastructure
Section titled “12. Skipping the boring infrastructure”DR drill never run, pen test never done, secrets rotated never, access review skipped. All fine until the day they aren’t, and then they’re catastrophic.
Refuse: the boring engineering work is the most-important work. Done quarterly, no exceptions.
What protects the business
Section titled “What protects the business”- Discipline in saying no to scope creep.
- Tight credit policy with frequent vintage review.
- Compliance first-class in every meeting.
- CA / Tally / embedded distribution as the differentiator.
- Repeat-borrower compounding — second loan more profitable than first.
- Lean operations — automation, not headcount, scales the business.
- Strong co-lending partner relationships — patient, long-term.
- Investor relationships — capital availability never the binding constraint.
The blueprint in this site is a starting point. Execution is the rest.