\n
\n<\/p>\n
Declining good borrowers, missing fraud, and watching competitors outgrow you \u2014 here’s how to know when your underwriting model has run its course.<\/em><\/p>\n Most lenders don’t set out to build a broken underwriting process. They build what works for the tools and data available at the time, and then they keep using it \u2014 long after the tools have improved and the data landscape has changed. The result is a model that made sense in 2010 but is quietly underperforming in 2025, leaving money on the table and risk on the books without anyone quite knowing why.<\/p>\n Cash flow underwriting isn’t a replacement for everything you’ve built. It’s an upgrade \u2014 a way to add a layer of real-world financial behavior data on top of your existing process and make every decision sharper. But knowing when to make that upgrade requires recognising the symptoms first. Here are five signs your lending process is overdue.<\/p>\n This is the most common \u2014 and most expensive \u2014 sign that something is wrong. If your underwriting model is declining a significant percentage of applications but your default rate isn’t impressively low, you’re getting the worst of both worlds: turning away creditworthy borrowers and<\/em> still approving some bad ones.<\/p>\n The reason is almost always the same: credit scores are a blunt instrument. They sort applicants into broad buckets \u2014 prime, near-prime, subprime \u2014 but they can’t distinguish the genuinely risky borrowers within each bucket from the ones who simply don’t have thick credit files. A thin-file applicant with consistent income and healthy savings is a very different risk from a thin-file applicant with erratic deposits and chronic overdrafts. Credit scores treat them the same. Cash flow underwriting doesn’t.<\/p>\n Research from the Consumer Financial Protection Bureau has found that adding cash flow data to underwriting models improves the prediction of late payments beyond what credit scores alone can achieve. That means you can approve more borrowers from the grey zone \u2014 not by relaxing standards, but by measuring risk more accurately. If your decline rate is high but your portfolio still underperforms your expectations, that gap is your opportunity.<\/p>\n The gig economy has reached what researchers now call “critical mass.” Over 70 million Americans \u2014 approximately 36% of the total US workforce \u2014 now participate in some form of freelance or contract work. Full-time independent workers more than doubled from 13.6 million in 2020 to 27.7 million in 2024. By 2027, freelancers are projected to represent over 50% of the US workforce.<\/p>\n Traditional underwriting was built for W-2 employees with predictable payroll deposits and verifiable employer relationships. That describes a shrinking share of the applicant pool. Self-employed borrowers have real income \u2014 the 4.7 million independent workers who earned over $100,000 in 2024 alone makes that clear \u2014 but their income doesn’t show up neatly in a credit file or on a standard pay stub.<\/p>\n If your underwriting team is spending increasing amounts of time on manual reviews for self-employed applicants \u2014 or if your approval rates for this segment are significantly lower than for W-2 borrowers \u2014 your process is telling you it wasn’t designed for the workforce you’re now serving. Cash flow underwriting solves this by going directly to the transaction record: the deposits either show up, consistently, or they don’t.<\/p>\n By 2027, freelancers are projected to represent more than 50% of the US workforce. Lenders whose underwriting models can’t handle non-W-2 income won’t just miss an opportunity \u2014 they’ll be locked out of the majority of the borrower market.<\/p>\n<\/blockquote>\n If you’ve added more verification steps but fraud losses keep rising, you’re fighting the wrong battle. According to Point Predictive’s 2025 Auto Lending Fraud Trends Report \u2014 which analysed fraud patterns across $4 trillion of submitted loan applications \u2014 first-party fraud accounts for 69% of the $9.2 billion in fraud risk exposure the auto lending industry faced in 2024. Income and employment misrepresentation alone accounted for $3.6\u20133.9 billion of that total.<\/p>\n The uncomfortable truth is that more document verification doesn’t solve income fraud. In 2024, 1 in 5 pay stubs submitted to auto lenders were forged. A 644% increase in AI and deepfake fraud discussions on criminal channels between 2023 and 2024 signals that fabrication is only getting easier and more convincing.<\/p>\n The only way to truly verify income is to bypass the documents and look at the money itself. Twelve to twenty-four months of transaction data from a borrower’s actual bank account is very hard to fabricate convincingly \u2014 the pattern of deposits, their timing, their consistency, their relationship to known income sources all either add up or they don’t.<\/p>\n Fintech lenders have been quietly building a structural advantage for years. By incorporating cash flow data into their underwriting models, they’ve been able to approve borrowers that traditional lenders decline, at rates that still reflect appropriate risk. A 2024 Datos Insights survey of 434 global consumer lending institutions found that cash flow underwriting adoption is accelerating rapidly across the industry.<\/p>\n Lenders who have already adopted cash flow underwriting are compounding their advantage: they’re building performance data, refining their models, and learning which cash flow signals predict default most reliably in their specific loan types and borrower profiles. Lenders who haven’t started yet are falling further behind with every quarter.<\/p>\n Manual review is expensive, slow, and doesn’t scale. If your underwriting team is regularly pulling additional documentation, calling employers, requesting multiple bank statement cycles, or escalating applications because the credit score alone doesn’t tell them enough, they’re compensating for a model that isn’t giving them what they need.<\/p>\n Cash flow underwriting automates the most labour-intensive part of this process. Instead of your team manually combing through bank statements and trying to categorise income sources, Kora’s platform does it automatically \u2014 identifying income deposits, assessing consistency, calculating effective debt service coverage, and flagging anomalies \u2014 and surfaces the results as a single score alongside supporting data. The same analysis that might take an underwriter an hour takes seconds.<\/p>\n If you recognise two or more of these signs in your current process, the question isn’t whether to add cash flow underwriting \u2014 it’s how quickly you can do it. At Kora, we’ve built cash flow underwriting infrastructure specifically for lenders who want to upgrade their process without rebuilding it. The Kora Score integrates directly into your existing loan origination system, adds transaction-level analysis to every application, and surfaces the results in a format your team can act on immediately.<\/p>\nSign #1: Your Decline Rate Is High, But Your Default Rate Isn’t Low Enough to Justify It<\/h2>\n
Sign #2: A Growing Portion of Your Applicants Are Self-Employed or Gig Workers<\/h2>\n
\n
Sign #3: Your Fraud Losses Are Climbing Despite Tighter Verification<\/h2>\n
Sign #4: You’re Losing Market Share to Fintechs and Alternative Lenders<\/h2>\n
Sign #5: Your Underwriters Are Spending Too Much Time on Manual Reviews<\/h2>\n
What to Do Next<\/h2>\n