The Data Confirms It: Traditional Credit Scoring Is Leaving Money on the Table
Lenders trust their credit data less than they used to. That is the finding at the center of a new report from Datos Insights, sponsored by Plaid. And it explains why more financial institutions are turning to consumer-permissioned data to fill the gaps.
The Trust Deficit
Traditional credit models are getting better at predicting outcomes for the people they can see. The problem is who they cannot see.
In a survey of 200 U.S. lending institutions, 75.5% reported improved predictive performance from traditional credit data. Yet only 52.5% said they feel more confident making lending decisions based on bureau data alone. Nearly a quarter, 24.5%, said they feel less confident than they did a year ago.
That gap is not a coincidence. It reflects what lenders see every day: entire populations of financially active, creditworthy people who traditional models cannot score.
The Applicant Pool Is Bigger Than Bureau Files Can Capture
Three numbers from the report stand out for anyone working in credit infrastructure:
- 47.5% of lenders report that 30% or more of their applicant pool is unscorable with traditional data alone
- 24.5% of U.S. adults earned some form of gig income in the past year, activity that most bureau-based models still cannot see
- 29.1% of consumers used buy now, pay later services in the past 12 months, often as a sign of tight cash flow rather than simple convenience
These are not edge cases. They are a quarter to nearly half of the market, depending on the institution. For financial institutions, that translates into real lending opportunities going unrecognized or unaddressed, not because the applicants are risky, but because the data used to evaluate them was never built to capture how they actually manage money.
Confidence Is Shifting Toward Alternative Data
The report tracks a real shift in sentiment. 74% of lenders say they are more confident in alternative credit data than they were a year ago, compared to 52.5% who feel more confident about traditional data alone. And it is not just sentiment. Among lenders who have adopted alternative data, 80% report improved portfolio performance, with 39% calling the improvement significant.
Bank transaction data leads current adoption, already in use at 56% of institutions. Utility, rental, and telco payment data are close behind, with consideration rates for each now exceeding current usage. That is a signal that adoption is about to accelerate.
Why This Matters for Bloom Credit
This is the exact gap Bloom+ was built to close.
Consumer-permissioned data lets financial institutions report account holders’ existing bill payments, including rent, utilities, and Telecom, to credit bureaus. It builds credit history from payments people are already making, without requiring new debt.
The results back up what the Datos Insights research points to industry-wide:
- Bloom+ enrollees start with an average credit score of 657
- Scores increase 16 to 23 points within 72 hours of enrollment
- 61% of enrolled members qualify for a new loan within 90 days
- Enrolled account holders show a 31% increase in monthly bill payments
- Retention sits at 98% or higher one year post-enrollment
The report’s core question is simple: can this person afford the payment right now, based on how they actually manage money? That is a forward-looking question traditional bureau data cannot answer on its own. Consumer-permissioned data gives financial institutions a real-time way to answer it.
The Institutions Moving First Will Own These Relationships
The report’s own recommendation is direct: institutions that integrate alternative data now are building a compounding advantage. Better data leads to better models, better decisions, and better portfolio performance, which in turn funds further investment. Institutions that wait are not standing still. They are falling behind lenders who already have a head start.
For credit unions and community banks serving members with thin files, gig income, or a history of missed opportunities under old scoring models, that head start is available today.
Source: Datos Insights, “Traditional Credit Models Lose Effectiveness in an Evolving Economic Environment,” sponsored by Plaid, June 2026.
