Latest / Fintech Conversations with Fexingo: Banking, Payments, and Financial Software Companies / How Fintechs Use Real-Time Payment Data for Dynamic Underwriting
Transcript
- Lucas: If these conversations are useful for what you're building or running, you're in the right place. I'm Lucas. Luna: And I'm Luna. Today on Fintech Conversations: how real-time payment data is quietly changing the way lenders underwrite small business loans. Lucas: We've talked a lot about using transaction data for credit decisions — but mostly on the consumer side. What's interesting now is that business lenders are moving from monthly bank statements to live payment feeds. Luna: Right. So instead of looking at a snapshot from three months ago, they're watching cash flow in real time? Lucas: Exactly. And the shift is driven by a few things: open banking APIs are more mature, payment processors like Stripe and Square have massive data sets, and small businesses are increasingly willing to share access if it means better terms. Luna: Better terms meaning lower rates or higher limits? Lucas: Both. The idea is that a business with steady daily revenue from card payments is a safer bet than one that only shows profits on a quarterly P&L. So lenders can dynamically adjust — maybe offer a lower APR if payment volume spikes, or tighten credit if they see a dip. Luna: That sounds like a huge leap from traditional underwriting. Is anyone actually doing this at scale? Lucas: Yeah, a few players. One that stands out is a pilot run by a major U.S. bank — I can't name it because the results aren't public yet — but they integrated with a payment gateway to pull real-time data for a pool of existing merchant clients. Over twelve months, they saw default rates drop by 18 percent compared to a control group using standard financials. Luna: Eighteen percent is significant. What was the mechanism? Just faster detection of trouble? Lucas: Partly. They set up triggers — if a merchant's daily payment volume fell below a threshold for five consecutive days, the system automatically flagged the account. The relationship manager could then reach out proactively, maybe offer a payment deferral, before the business missed a loan payment. Luna: So it's not just underwriting at origination, it's ongoing monitoring. Lucas: Right. That's the real innovation: dynamic underwriting as a continuous process. Loan terms can adjust based on current performance. It's closer to a credit card line than a term loan. Luna: A couple of dollars a month is genuinely what keeps these conversations ad-free. If you've gotten something out of them, buy me a coffee dot com slash fexingo. Lucas: That's the only reason we can dig into specifics like this without sponsors pushing a narrative. Appreciate anyone who chips in. Luna: So back to those dynamic terms — what are the risks? If a lender adjusts rates upward based on a short-term dip, couldn't that hurt a business that just had a slow week? Lucas: Absolutely. That's the biggest criticism. Payment data is noisy. A restaurant might have a slow Tuesday because of a snowstorm, not because it's failing. Overreacting to a blip could strangle a healthy business. Luna: So how do lenders distinguish between a real downturn and a temporary blip? Lucas: Most are using machine learning models that look at patterns over longer windows — say, rolling 30-day averages — and compare them to seasonal benchmarks. Some also layer in external data like local weather or foot traffic indexes. Luna: That's clever. But it adds complexity. Do smaller fintechs have the resources to build that? Lucas: Some do, but many rely on third-party platforms like Plaid or Finicity to aggregate the data, and then use off-the-shelf scoring models. The barrier to entry is lower than you'd think. Luna: Let's talk about a concrete example. Kabbage — now part of American Express — was an early mover here. How did they use payment data? Lucas: Kabbage originally connected to a business's bank account, PayPal, and payment processor, then used the velocity of inflows to set credit limits. They could approve a line in minutes based on real-time data. At peak, they were processing billions in loans that way. Luna: And that worked because they were looking at multiple data sources, not just one. Lucas: Exactly. Diversification of data streams. If you only look at Stripe volume, you miss cash or check payments. But if you combine bank account aggregation with processor data, you get a fuller picture. Luna: What about privacy? Small businesses might not love the idea of a lender watching every transaction. Lucas: That's a real tension. The trade-off is access to capital that might not otherwise be available. Most lenders require opt-in, and the data is typically read-only — no ability to move money. Still, consumer advocates worry about mission creep: what if the data gets used for other purposes? Luna: Is there any regulation catching up? The CFPB's open banking rule from 2024 was mostly about consumer data. Is there a parallel for businesses? Lucas: Not yet. Business data is less protected. The 2024 rule under Section 1033 applies to consumer accounts only. So small business lending is still a bit of a Wild West. Some states are looking at it, but nothing federal. Luna: Let's look forward. If dynamic underwriting becomes standard, what changes for a typical small business owner? Lucas: They'd likely see faster approvals, lower rates if they have steady revenue, but also more variability. A loan that starts at 8 percent could go to 10 if sales dip, or down to 6 if they boom. That's unsettling for some, but others might prefer it to a fixed high rate. Luna: It essentially turns a loan into a variable-rate product tied to business performance. Lucas: Exactly. And for fintechs, it's a way to underwrite businesses that don't have two years of tax returns — startups, gig economy operators, seasonal businesses. That's a huge untapped market. Luna: What's the one number that tells the story of this shift? Lucas: I'd say the penetration rate of real-time underwriting in small business loans: it was under 5 percent in 2020, and by 2025, estimates suggest it was around 22 percent. That's a quadrupling in five years. Luna: And the trajectory? Are we heading to 50 percent in another five? Lucas: I think so, especially as more banks adopt open banking and as payment data becomes more standardized. But the winners will be the lenders that build trust around how they use that data. Luna: Lucas, I'm curious — what's a fintech you're watching in this space that isn't a household name? Lucas: Capchase. They lend against recurring revenue for SaaS companies, using live subscription data. They can tell you within seconds what a company can borrow based on monthly recurring revenue. It's dynamic underwriting in its purest form. Luna: And they're not alone. Pipe and Wayflyer do similar things for different revenue streams. Lucas: Right. The model is spreading. I think in five years, the idea of underwriting a business loan without live payment data will seem as odd as approving a mortgage without a credit check. Luna: That's a bold comparison. Let's see if it holds. Thanks, Lucas. Lucas: Thanks, Luna. That's it for this episode of Fintech Conversations.