Credit decisions that expand access without expanding risk
Alternative data credit scoring lifting approvals 25% while holding defaults below industry average.
Millions of creditworthy people are invisible to traditional scoring—alternative data changes that.
The Challenge
A digital lender was rejecting 40% of applicants who could safely receive credit because traditional FICO scores could not assess thin-file consumers. The lender served a demographic with high proportions of recent immigrants, gig workers, and young adults—populations with limited bureau history but verifiable financial behavior. Competitors using alternative data were capturing market share, and regulators expected the lender to demonstrate fair access efforts.
The Innovoco Solution
We built an alternative credit risk model incorporating utility payments, rent history, bank transaction cash flow patterns, education, and employment signals—with Fair Lending compliance baked into the model development process, including ECOA and disparate impact testing at every stage.
Phase 1 — Feature engineering and compliance framework
Identified and validated alternative data sources (utility, rent, bank transaction, employment). Established a Fair Lending testing framework with legal counsel: every candidate feature tested for disparate impact before inclusion. Built baseline models with and without alternative data to quantify approval lift and default impact.
Phase 2 — Production deployment with monitoring
Deployed the alternative scoring model alongside traditional scoring. Applicants scored by both models; the alternative model expands approvals for thin-file applicants who would otherwise be declined. Ongoing monitoring tracks default rates, approval demographics, and model drift by segment.
Key implementations
Alternative data integration
Utility payment history, rent reporting, bank transaction cash flow, and employment verification integrated with consumer consent and data quality validation.
Fair Lending compliance
Disparate impact testing on every model feature and the composite score. Adverse action reasons generated for every decline. Regular audits with legal and compliance teams.
Explainable scoring
Transparent feature attribution for every credit decision, enabling both regulatory explanation and applicant-facing adverse action notices.
Dual-model architecture
Traditional and alternative models run in parallel; the alternative model activates only for applicants where bureau data is insufficient, avoiding disruption to existing credit policy.
Performance monitoring by segment
Default rates, vintage curves, and approval demographics tracked by score tier and data source to detect drift and ensure continued Fair Lending compliance.
Technical Innovation
A modular feature pipeline allows new alternative data sources to be tested for predictive power and fair lending impact independently before inclusion in the production model. The compliance testing framework runs automatically with each model update, preventing regressions.
Impact
- 25% increase in approval rates for thin-file applicants without relaxing credit standards.
- 5.0% default rate on alternative-scored loans versus 6.5% industry average on comparable segments.
- 100% ECOA compliance with documented disparate impact testing for every model version.
- Time-to-decision unchanged—alternative scoring adds no latency to the application experience.
The lender expanded access to creditworthy consumers that traditional scoring overlooked—while demonstrating to regulators that alternative data improved both access and risk performance.
Explore this outcome on your stack
We map scope, guardrails, and rollout to your data boundaries and teams—practical next steps, not a generic slide deck.
60 min · Free · No obligation
More case studies
Explore adjacent patterns from the same production playbook.
AI-Powered Business Intelligence
Ask questions in plain language, get governed answers with charts — no analysts, no ticket queue.
Read more
Customer support & voice
Triage, sentiment-aware routing, CRM-grounded answers at scale.
Read more
Regulated onboarding (KYC / AML)
Document intelligence, exception queues, immutable evidence for auditors.
Read more