Apollo’s almanac to evaluate a Digital Lending NBFC

A few months ago, Our team walked into the office of a digital lending NBFC that, at first glance, seemed like the poster child for success. The energy in the room was infectious — employees chatted over free lunches, and the walls were plastered with motivational quotes about financial inclusion and technology. Their app branding was slick, their marketing campaigns were everywhere, and the founders confidently spoke about their growing profitability. It all looked perfect, almost too perfect.

But as we dug deeper — reviewing their static pool data, portfolio delinquency reports, and provisioning policies — we realized that the numbers were telling a very different story. Their provisioning levels were far below what any prudent lender would maintain, and their so-called profitability hinged on accounting adjustments that masked the real risks in their book. This was not uncommon. In the digital lending industry, we’ve seen many examples of numbers dressed up to look better than they truly are.

Over the past few years, Apollo has evaluated more than 27 NBFCs. We’ve conducted ground visits, met founders face-to-face, analyzed financials line-by-line, and built our own benchmark ratios based on lessons from both the successes and failures we’ve observed. Through this journey, we’ve developed a framework for evaluating NBFCs that goes beyond surface-level metrics. In this blog, I’ll share this framework — rich with examples, anecdotes, and mental models — in the hope that it will help you distinguish a truly strong NBFC from one that’s simply riding on hype.

1. Profitability: The Mirage of Adjusted Numbers

Since 2022, profitability has become the metric that every fintech founder wants to showcase. Nearly 90% of NBFCs we reviewed in the past year claim they are profitable. But profitability in lending isn’t as straightforward as it looks. We at Apollo have developed a measure we call “Apollo Profit” — a recalibrated view of profits that adjusts for conservative provisioning, hidden expenses, and one-off accounting maneuvers.

The Provisioning Game

Provisioning is where many NBFCs play the optics game. Think of provisioning as a financial safety net — the amount set aside for anticipated loan losses. RBI’s minimum provisioning norms are relatively low, sometimes just 0.5% for certain loan buckets. But digital lenders, with their predominantly unsecured loan portfolios, face delinquency rates that can be 10x higher than those in traditional secured lending.

Here’s an example: A company might have a 10% portfolio at risk (PAR 30+), but it only books 0.5% provisioning to meet the bare regulatory minimum. On paper, their profits look stellar. A more conservative NBFC would provision 5-10% against the same PAR, which paints a much truer picture of potential losses but lowers apparent profits. We’ve also seen companies intentionally over-provision during good years to smooth future earnings — a strategy that can mislead if not analyzed carefully.

The timing of provisioning is equally important. A fast-growing NBFC can hide past delinquency issues by rapidly expanding its AUM (Assets Under Management) and delaying provisions. We always ask: Is this company provisioning at the same pace as its AUM growth? If not, the profitability might be an illusion.

Anecdote: We came across one NBFC where the legal language in the loan agreement — combined with how their LMS was structured — effectively deferred the appearance of delinquencies. Instead of defaults showing up during the active repayment period (say in year 1 or 2 of a 3-year loan), most delays were pushed to the end of the tenure. This wasn’t a data glitch; it was a product intentionally designed to showcase delinquency much later than it would otherwise appear. At Apollo, we read the fine print of loan agreements, review the LMS setup, and understand product structures in depth — because only by triangulating across all three can we determine if delayed defaults are a design feature, not a bug.

2. The Parent Entity Trick: Where Costs Hide

Around 60% of NBFCs we’ve assessed operate with a “dual-entity” structure — a parent tech company and a lending subsidiary. This setup allows founders to shift expenses strategically, often parking significant costs like CXO salaries, tech development, or marketing under the parent entity. The NBFC’s books, as a result, look healthier than they really are.

The employee benefit expense/disbursement ratio in the industry ranges between 1.5–3%. Of course, there could be reasons for being an outlier — but that’s what you need to be able to justify. Are you a truly efficient tech NBFC, or are you simply pushing expenses into the parent entity’s books? Is your CEO’s remuneration hidden under a different head altogether in the financial statement, or is the CEO not drawing a justifiable salary today to chase a stronger bottom line? They could also be investing ahead — building a strong team with the expectation of scaling disbursements in the future. What matters is whether this story holds up when viewed alongside the rest of the metrics — and whether it continues to do so as the business scales.

3. Debt-to-Equity: The Power and Peril of Leverage

Lending is, at its core, a leverage-driven business. RBI allows NBFCs a debt-to-equity ratio of up to 7:1, but in practice, we rarely see digital NBFCs exceed 3:1 because of their limited track records and relatively small net worths.

The key is to look at consolidated leverage. An NBFC might show a conservative 2:1 ratio, but its parent entity could be raising expensive venture debt to fund its growth. Conversely, an NBFC may seem over-leveraged, but with strong parent-level equity, the overall risk might be acceptable.

Observation: In one instance, we found that an NBFC was borrowing at 16-18% through its parent company and internally funneling the capital to its lending arm at a much lower cost. The NBFC’s cost of capital looked attractive, but the actual funding structure was fragile. When we asked the founders how long this cross-entity leverage could be sustained, they admitted it was not scalable.

Excessive leverage on an unproven book is risky. Any regulatory tightening or unexpected spike in defaults can send a highly leveraged NBFC into a downward spiral. Sustainable growth means balancing the need for leverage with portfolio stability and equity backing.

4. Capital Adequacy (CRAR): The Safety Cushion

CRAR (Capital to Risk-Weighted Assets Ratio) tells you how much shock an NBFC can absorb before it becomes insolvent. RBI mandates a minimum CRAR of 15%, but most well-run digital NBFCs operate at 20-25%. We view sudden drops in CRAR as a warning sign.

For instance, if a company’s AUM grows aggressively while its own capital (equity + retained earnings) lags behind, the CRAR will drop. This is often a sign of growth being fueled by external debt rather than sustainable internal profitability. We prefer NBFCs that grow steadily, maintaining a balance between debt, equity, and retained earnings.

Case Study: Imagine an NBFC that starts out strong — a CRAR of 35% in its early days, signaling a solid capital cushion. But by year three, its AUM grows aggressively, almost entirely fueled by debt. In just one year, that CRAR drops to 17%. On its own, the drop could be explained — growth often causes some dilution in ratios. But here’s where it becomes a concern: profits haven’t kept pace, equity hasn’t been raised, and internal accruals are thin. That combination — rapid debt-led growth without a matching uptick in profitability or capital infusion — is exactly the kind of pattern we flag as high-risk. It’s a growth story that could buckle under its own weight.

5. Portfolio Quality: The True North

Portfolio quality is the single most important factor in determining whether an NBFC will survive and scale. A high-quality book can withstand shocks and attract better lenders. A weak book, no matter how well marketed, is a ticking time bomb.

We analyze portfolio quality through cohort and static pool analysis.

• Static Pool: This shows the worst delinquency levels that each vintage of loans (say, Jan 2023 disbursements) has reached over its lifecycle, without being diluted by new disbursements.

• PAR (Portfolio at Risk): We look at PAR 30, PAR 60, and PAR 90 to get a realistic picture. Gross NPA or Net NPA numbers, which often look good due to early write-offs (industry average is between 90-120 days), don’t tell the whole story.

Example: A digital NBFC we met was proud of its <1% GNPA. However, their static pool data revealed something else: their Jan, Feb, and March 2022 cohorts had PAR 30 rates of 8-12%. The write-offs were simply happening quickly — usually 120 days — so NPA ratios stayed artificially low. When we ran the numbers using consistent aging methods, the true health of the book became clear.

6. DNA of the Company: Is It Truly Tech-First?

Every NBFC claims to be a “tech NBFC,” but only about 35% of those we’ve evaluated genuinely use machine learning models or algorithmic underwriting. The rest rely on business rule engines (BREs), which are just automated versions of traditional underwriting.

How do we test this?

• Headcount: If you claim to be fully tech-driven but have 500 people managing collections, something doesn’t add up.

• Expense Ratios: High collection costs often indicate a manual, not tech-first, approach.

• Operational Efficiency: A genuine tech NBFC should show strong productivity ratios and scalable unit economics.

On-the-ground Example: One of the most impressive NBFCs we evaluated had under 50 employees, yet was disbursing tens of crores monthly. They had automated customer onboarding, underwriting via an internal ML engine, and collections through a self-service repayment app and WhatsApp bots. Their collection expense was 0.6% of disbursement — nearly half the industry median. The contrast with some of their peers — who had hundreds of field agents and higher cost ratios — was striking.

7. Promoter Skin in the Game

Lending is as much about people as it is about numbers. We prefer NBFCs where founders have significant equity stakes (30-40%) and are clearly aligned with the company’s long-term success.

Fact: More than 60% of digital NBFC founders we’ve reviewed have diluted their stakes to below 30% in just 5-6 years. While this may be due to aggressive fundraising, it often signals misaligned incentives. Compare this to Uday Kotak, who still owns ~ 25% of Kotak Mahindra Bank after decades of building.

Excessive dilution at an early stage suggests either high burn, unproven models, or frequent pivots. We treat such cases with caution.

Example: One NBFC had raised three rounds of equity in four years. Their founder’s stake had fallen to just 12%. When asked about it, he candidly said, “We needed to raise to survive. Our first two product lines didn’t work, and we pivoted twice.” While we appreciated the honesty, the frequent pivots and heavy dilution signaled a fragile foundation. We decided to pass on the deal.

Lessons from Evaluating 27 NBFCs

• Lesson 1: Profitability without conservative provisioning is meaningless.

• Lesson 2: Always adjust for hidden costs across entities.

• Lesson 3: AUM growth without a strong CRAR or static pool performance is a red flag.

• Lesson 4: Tech claims must be validated through operational data, not just pitch decks.

• Lesson 5: Dilution tells a story. Read it.

The Final Word

Digital lending is a marathon, not a sprint. As Jeff Bezos famously says, “All overnight success takes about 10 years.” In lending, slow and steady often wins the race. The best NBFCs aren’t those that shout about rapid growth or short-term profitability, but those that quietly master the fundamentals: portfolio quality, prudent provisioning, sustainable leverage, and aligned leadership.

Our advice? Look for truth, not trends. A well-run NBFC is built like a fortress, brick by brick, over time. The DNA of the company — its leadership, its book quality, and its culture of transparency — will always tell you if it’s built for the long game.