When Is It Worth Lending to Marginal Customers?

Brandon Homuth
Feb 23
3 min read

Every lender faces a version of this question:

Should we approve customers who are barely profitable today, hoping they’ll become valuable later?

These “marginal” users sit right on the edge of profitability — their expected NPV is close to zero. They’re the hardest to classify, yet they often represent the biggest opportunity for learning and growth.

Handled well, they help you expand your frontier, improve models, and capture market share. Handled poorly, they drain liquidity, trigger covenants, and distort your view of portfolio health.

So when is it worth lending to them — and when is it better to wait?

The Case for Approving Marginal Customers

If you had infinite capital and time, the answer would be simple: approve everyone with NPV > 0. Each positive-NPV customer adds value, even if it’s small. The trick is that “positive NPV” depends on your assumptions about losses, costs, and behavior — and those assumptions are rarely perfect.

That’s where marginal users become useful. They’re the test bed for model learning and market expansion.

When you approve them intentionally, you gain data that helps you:

Validate your model at the boundary of its decision space.
Quantify differences between channels or segments.
Learn how customers evolve as your product improves.
Identify profitable niches that traditional models overlook.

In other words, approving some marginal customers is a way to buy information — and in early-stage or rapidly evolving portfolios, that information can be more valuable than short-term profit.

The Case for Saying No

But not all learning is worth the tuition.

When liquidity is tight, debt covenants are binding, or your cost of equity is high, even slightly negative NPV decisions compound into real risk.

Worse, a poor early experience can “spoil” a customer segment. If you bring them in too soon — before your product, servicing, or pricing is ready — you may burn trust and make re-acquisition harder later.

The opportunity cost is also nontrivial: each dollar spent approving marginal users could have gone to stronger cohorts or higher-ROI channels.

So while learning is valuable, the decision must reflect your constraints and stage. The earlier and better-capitalized you are, the more risk you can take for information. The later and more leveraged you are, the more discipline you need to preserve capital and covenant headroom.

Modeling the Trade-Off

Quantitatively, this trade-off can be expressed as a simple decision framework:

Baseline NPV. Estimate customer-level or segment-level NPV under current assumptions.
Stressed NPV. Recalculate assuming losses or revenue deteriorate by X% (e.g., 20–50%). If stressed NPV remains positive, approve.
Information Value. Assign value to learning — for example, the expected lift in future NPV from improved model calibration.
Leakage Risk. Estimate how much of the “parked” segment you can realistically reacquire later (often only 50–70%).

The framework forces a clear question:

Is the expected information gain and future retention worth more than the short-term capital cost and leakage risk?

In many cases, that answer depends on how confident you are that your product and credit model will improve — and how fast.

How Leading Lenders Manage the Margins

In our work with fintechs and lenders, we see three effective approaches:

Dynamic scorecuts by channel maturity. Approve down to NPV ≈ 0 in testing channels (to learn), and keep stricter thresholds in mature channels (to earn).
Segmented experimentation. Define clear, bounded “learning zones” — e.g., 5% of approvals allocated to marginal users — and monitor NPV, loss rates, and learning impact separately.
Re-solicit with intention. For deferred segments, plan a specific re-entry path: track them, improve the product, and come back when your economics improve. Don’t rely on chance rediscovery.

These practices make marginal approvals a strategic investment, not a random leak in credit discipline.

The Bottom Line

The biggest mistake isn’t approving marginal customers — it’s doing it unintentionally.

If you can quantify the trade-off between short-term profit, long-term learning, and capital constraints, you can turn “marginal” users into a controlled growth lever instead of a hidden liability.

The goal isn’t to avoid risk. It’s to underwrite risk with purpose.