How storefront and online lenders can profitably expand access to underbanked customers — and why BloomGrade makes it low-risk and low-cost.
The next meaningful growth opportunity for consumer lenders won’t come from squeezing more margin from prime borrowers.
It’s in the millions of people left out or poorly served by traditional credit scores.
These “thin-file” or “unscorable” consumers (and the broader underbanked population they often belong to) represent both sizable demand and potentially lower-than-expected credit risk when measured the modern way.
Here’s the research, explaining how BloomGrade lets lenders find and filter these customers without adding per-lead data costs, and gives practical steps storefront and online lenders can take today.
The market: large, growing, and underserved
Several recent analyses agree on two points: a) a meaningful share of U.S. adults have thin or unscorable credit files, and b) a still-larger group is underbanked and relies on alternative financial services. The Federal Deposit Insurance Corporation’s household survey shows about 14.2% of U.S. households were underbanked in 2023 (≈19 million households).
Research that updates estimates about “credit invisibles” and thin/stale files finds that millions more consumers fall into categories lacking traditionally scored credit records — creating an addressable market that mainstream models systematically exclude. The New York Fed and CFPB analyses describe between ~8–19 million additional consumers with unscorable or thin credit files depending on definitions and year.
Why that matters: these are real people with rent, utilities, paycheck behavior, and payment patterns that traditional FICO-style models don’t fully capture — but that predictive analytics and alternative signals can.
Risk reality: thin files ≠ high defaults
The academic and policy literature shows alternative data and modern ML can surface predictive signals not in traditional bureau histories, allowing some thin-file consumers to be safely extended credit at acceptable loss rates. Fintech research by Jagtiani and others finds alternative data + machine learning can both expand access and lower underwriting costs — and in some settings enable better risk discrimination than legacy methods.
Put simply: many thin-file borrowers are creditworthy but invisible to legacy scores. If lenders ignore them, they forfeit fee and interest income — and miss opportunities for high-LTV, low-cost customer acquisition.
The blocker: data cost and operational friction
Why haven’t most storefront and online lenders moved aggressively into this segment? Two common barriers:
- Per-app data costs and integration friction. Pulling layered alternative data sources per applicant (rent, telco, bank-transaction enrichments, psychometrics) can add vendor fees and engineering work that shrink margins.
- Regulatory and fair-lending complexity. Using new signals requires careful model governance and documentation.
That’s where a scoring layer that works without per-lead third-party pulls becomes a game changer.
How BloomGrade unlocks the market (without per-lead data costs)
BloomGrade is a suite of scores designed to let lenders identify high-potential thin-file leads using existing application and lightly enriched data, not expensive per-lead vendor calls.
Here’s the practical mechanics and business value:
- Modeling from minimal inputs: Bloom learns from patterns in the application itself (employment, income cadence, contact stability, device & behavioral metadata where available) plus historical loan outcomes to estimate probability of default. Because the model is trained to extract predictive power from information lenders already collect, lenders avoid frequent external data pulls and vendor fees. (This is the “no extra data cost” advantage.)
- BloomGrade = high-precision lead filtering: BloomGrade maps the scores into discrete lead buckets (example: Grade A/B/C) so lenders can apply different decisions and price points. A storefront lender can route Grade A applicants to instant in-store approvals with a favorable APR and cross-sell; Grade B can go to thin-file monitoring or a small starter product; Grade C can receive prescreened marketing rather than an underwriting attempt. This preserves margins by allocating cost-intensive manual review or external verification only to the most promising cases.
- Look-Alike Scoring Capability for Thin-File Expansion: Many lenders lack the internal data science capability to build reliable “look-alike” models—models that allow them to infer likely borrower behavior without deep individual credit histories. Bloom fills that gap. By training across large datasets of borrower attributes and loan outcomes, Bloom can identify applicants who resemble historically successful borrowers—even when traditional bureau data is sparse or unavailable. This enables lenders to confidently expand into thin-file segments without adding new data costs, building complex in-house models, or increasing uncontrolled risk.
- Calibration to loss tolerance & portfolio strategy: Because BloomGrade is a flexible filter, lenders control acceptance rate vs. expected loss tradeoffs. For example, an online lender focused on growth can accept a slightly lower threshold during a campaign and reprice; a storefront lender focused on profitability can set tighter grade cutoffs for funded loans.
- Operational simplicity for storefront teams: No complex vendor dashboard or per-lead decisioning—Bloom integrates with existing origination flows and returns a grade/score that staff can act on instantly, enabling consistent decisions across branches.
What BloomGrade Actually Solves
BloomGrade converts a score into simple action buckets.
Grades (A/B/C or similar) give lenders instant segmentation for approvals, product fit (starter/secured/standard), pricing tiers, manual review routing, and consistent branch decisioning.
Bloom’s scores predict default risk from the data you already collect.
It is a machine-learning PD (probability-of-default) score trained to extract maximum signal from application data, employment & income cadence, stability patterns, behavioral metadata (when available), and historical outcomes — without per-lead alternative data pulls.
In plain terms, Bloom helps lenders:
- Rank default risk
- Identify good borrowers they’re currently rejecting
- Improve conversion on thin-file applicants
- Segment borrowers into the right products and pricing
- Control losses while expanding access
- Reduce or eliminate per-lead data spend
Concrete tactics for storefront and online lenders
- Prescreen to uncover credit-invisible prospects
- Use BloomGrade as a prescreen in digital lead forms and branch check-ins. Rather than rejecting “no bureau match,” prescreen and present tailored offers (small starter loans, secured products, or educational enrollment).
- Result: higher conversion from walk-ins and low-intent digital visitors, with lower per-lead acquisition costs.
- Risk-based pricing and staged credit
- Offer starter loan products (short term, small principal) for Grade B customers with automatic re-evaluation and fast credit line expansion for on-time behavior. This drives lifetime value and controls initial exposure.
- Cross-sell non-credit products
- For thin-file borrowers, monetize by cross-selling payments, savings, and bill-pay tools that both improve retention and generate fee revenue while capturing more signal for future underwriting.
- Branch playbook
- Train staff to use BloomGrade decisions (approve, refer, educate) rather than relying on manual gut calls. Consistency reduces operational risk and supports compliance documentation.
- Compliance & documentation
- Maintain model governance records for Bloom Score decisions. Document input fields used, explainability summaries, and adverse action workflows where required.
Why this is profitable
A practical framework:
- More approvals → Bloom finds good borrowers you’re currently rejecting.
- Lower losses → PD scoring & grade thresholds align with your risk appetite.
- Lower acquisition cost → thin-file borrowers are less competitive to acquire.
- Lower data spend → no stacked per-app vendor fees.
- Better lifetime value → staged credit growth and cross-sell opportunities.
Summary: the market is ready — and Bloom makes it accessible
Thin-file borrowers represent a large, underserved, and mis-scored group that traditional models overlook. The lenders who win this segment first will gain:
- high-yield customers
- strong unit economics
- durable competitive advantage
Bloom gives lenders the ability to rank default risk, improve conversion, and segment/prioritize applicants — all without the cost burden of alternative data.
Found this article insightful? Share on LinkedIn.
DISCLAIMER: This content is informational only. Results vary by implementation. Lenders remain responsible for fair lending compliance and all regulatory requirements.