Why FICO Fails Small Businesses And What Smaller Lenders Are Missing Because of It
The scoring model built for consumers is quietly rejecting your most creditworthy SME borrowers
Small lenders know (or should know) their respective markets better than any algorithm. They know which restaurant owner has been packing the house for a decade, which contractor never misses a payroll, which wholesaler has been growing steadily for five years. And yet their own underwriting systems routinely decline these borrowers.
It is not a willingness problem. And It’s not even just a data problem. It’s the problem of using a scoring model designed for individual consumers to evaluate the creditworthiness of operating businesses.
The Model Wasn't Built for This
FICO was originally developed to assess individual consumer credit risk: personal credit cards, auto loans, mortgages. It works well for what it was designed to do. But small businesses are not consumers. Their financial profiles are structurally different. Revenues are variable and seasonal, personal and business finances are often intertwined, and the most important signal, cash flow, does not appear anywhere in a traditional credit report.
Research from Altman and Sabato (2007) (yes, almost 20 years ago) found that generic scoring models produce a 25.81% false rejection rate when applied to SME borrowers, meaning roughly one in four creditworthy small businesses is declined. That is not a rounding error. That is a structural feature of applying the wrong tool to the wrong problem. Even today. Especially today.
The Data Degradation Problem
The problem has gotten worse, not better. Traditional credit data is actively degrading as an input. Since 2020, more than 50% of credit card issuers have stopped reporting payment data. BNPL credit, which grew 8x in volume, is almost entirely unreported. The credit file that a scoring model is analyzing today contains less useful information than it did five years ago.
The market has noticed. A 2024 Forbes survey found that 60% of lenders report less confidence in traditional credit scores as a standalone indicator, while 86% report increased confidence in alternative data year-over-year. This is not a fringe view. It is the emerging consensus among the institutions doing the most lending.
What the Data Actually Misses
The information that best predicts whether a small business will repay a loan is not in its credit file. It is in its bank account. Cash flow patterns, transaction frequency, deposit stability, NSF frequency, even payroll consistency, among others, all provide real-time, forward-looking signals about a business's ability to service debt. Signals that are invisible to a FICO-based model.
Research from FinRegLab found that cash flow data provides a more detailed and timely picture of creditworthiness than traditional credit reports, particularly for startups and thin-file borrowers who simply do not have enough credit history to generate a meaningful score. This is not a small population. There are an estimated 26 million credit-invisible Americans and 19 million unscorable individuals in the US today.
The Business Cost of a False Rejection
Every false rejection has a dollar value. A creditworthy SME borrower denied a $100,000 loan generates zero net interest income, no fee revenue, and often exits the banking relationship entirely, taking their deposits and future business with them. Multiply that across a loan portfolio and the cost is not abstract.
The flip side is equally important: alternative data does not just reduce false rejections, it does so without increasing default risk. Research across multiple lenders using cash flow-based underwriting has shown approval rate improvements of 30-60% with equal or lower portfolio default rates. The borrowers being surfaced by better data are not riskier, but simply invisible to models that were not built to see them.
What This Means for Smaller Lenders
Community banks are uniquely positioned to benefit from this shift. They already have the relationship knowledge. They already have the customer trust. What they have historically lacked is a systematic, data-driven way to quantify what their loan officers already know intuitively about a borrower's financial health.
The good news is that the infrastructure to do this now exists and is deployable even for smaller lenders: no platform migration, no data science team, nor an enterprise-level budget. The lenders that move first on this will not just recover lost approvals. They will create a sustainable competitive advantage in the SME segment that fintechs, for all their speed, cannot replicate without local relationships.
