Latest / Fintech Conversations with Fexingo: Banking, Payments, and Financial Software Companies / Fintechs Use Real-Time Cash Flow Data for SMB Lending
Transcript
- Lucas: You know how a small business owner walks into a bank for a loan? They bring two years of tax returns, profit and loss statements, maybe a business plan. The bank takes weeks to decide. And by then, the opportunity — a new piece of equipment, a big order from a client — might be gone. Luna: It's absurd when you think about it. Banks are basically looking at a rearview mirror while the business is driving forward. Lucas: Exactly. And that's the gap fintechs are trying to close with what's called cash flow underwriting. Instead of relying on static documents from last year, they plug directly into the business's bank account or accounting software — think Plaid, or Xero, or QuickBooks — and look at real-time inflows and outflows. Luna: So they're not asking 'how profitable were you in 2024?' They're asking 'how much money is hitting your account every week right now?' Lucas: Exactly. And they can see patterns — does revenue dip every February and spike in November? That's seasonal, not risky. Do they have consistent repeat customers paying via invoice? That's predictable revenue. The algorithm learns what a healthy cash flow signature looks like for that industry and that business size. Luna: I've seen this with a company called Pipe. They let SaaS businesses borrow against their recurring subscription revenue in real time. You connect your Stripe account, they see your monthly recurring revenue trending, and they advance you a lump sum — often within hours. Lucas: Pipe's a great example. They're essentially turning a subscription stream into a loan without the founder giving up equity or going through a traditional credit check. The underwriting is entirely based on the cash flow data from Stripe. And it's not just SaaS — you're seeing this for e-commerce businesses on Shopify, for freelancers with steady invoice cycles. Luna: What about the banks' argument that traditional underwriting is more rigorous? They say cash flow data can be manipulated — a business could temporarily inflate revenue by running up credit card sales or taking a short-term loan. Lucas: It's a fair concern. But fintechs argue that because they're looking at granular transaction data — sometimes daily — they can spot those anomalies. If a sudden spike comes from a single large deposit that looks like a loan, the algorithm flags it. They're not just looking at the balance, they're categorizing every transaction. Luna: So it's more like a credit analyst that never sleeps, constantly updating the risk profile. Lucas: Exactly. And that opens up credit for businesses that would be invisible to traditional credit bureaus. The Small Business Administration says about 40 percent of small businesses don't have a traditional credit score — they're too new, or they operate mostly in cash, or they've never taken a loan before. Cash flow underwriting can still serve them. Luna: But it also creates a new kind of dependency. If a lender has direct read-only access to your bank account, they see every late payment, every bounced check, every dip in revenue. That could make them pull the line of credit just when you need it most. Lucas: That's the dark side. Some fintech lenders have clauses that let them recalculate your credit limit automatically based on trailing 30-day cash flow. If you have a slow month — say a big client pays late — your available credit could shrink overnight. That's not a risk you'd have with a traditional term loan. Luna: It's like algorithmic margin calls for Main Street. And there's a fairness question too: if the algorithm is trained on data from businesses that already have bank accounts and accounting software, it might systematically under-serve businesses that operate more informally — like many minority-owned businesses. Lucas: That's a real blind spot. And it's something regulators are starting to look at. The CFPB has signaled interest in how alternative data is used in lending, and there's a push for what's called 'algorithmic fairness' — testing whether cash flow models produce disparate impact across demographic groups. Luna: So the technology is powerful, but it's not a silver bullet. It can expand access, but it can also introduce new forms of volatility and exclusion. Lucas: Right. And the speed cuts both ways. A loan can be approved in hours, but the decision is also reversible in hours if cash flow dips. That makes it great for short-term working capital, but less suited for long-term investments like buying a building or funding R&D. Luna: Which brings us back to Pipe. They recently launched a product called Pipe 2.0 that lets companies borrow for longer terms — up to 12 months — by blending real-time cash flow data with a forward-looking forecast based on churn rates and expansion revenue. Lucas: So they're layering predictive analytics on top of the real-time data. Instead of just looking at what happened last month, they're projecting what's likely to happen next quarter based on the cohort's retention patterns. That's a significant evolution. Luna: It also means the underwriting model itself becomes a competitive moat. If you have better data connections and better predictive models, you can offer better rates and faster decisions than the next fintech. Lucas: Which is why you're seeing incumbents like JPMorgan and Bank of America investing in their own real-time cash flow platforms. They can't afford to let fintechs own this data layer. But they have legacy systems to integrate, which is harder than building from scratch. Luna: That's where companies like Plaid and Finicity come in — they're the plumbing that connects banks to fintechs. And they're becoming critical infrastructure. Lucas: Absolutely. And the regulatory landscape is shifting too. The UK already has open banking mandates that force the big banks to share transaction data with authorized third parties. The US is moving more slowly, but the CFPB's Section 1033 rule — expected later this year — would require banks to let consumers share their data with fintechs on request. Luna: That would be a game-changer. It would make cash flow underwriting much more widespread, because fintechs wouldn't have to rely on screen scraping or individual partnerships. They'd have standardized API access. Lucas: Right. And it would also give the consumer more control over who sees their data, which addresses some of the privacy concerns. But it also raises the stakes for getting the models right — because if every lender is using the same data sources, you could end up with a monoculture of credit decisions. Luna: That's a fascinating point. If all lenders see the same cash flow dip at the same time, they could all cut credit simultaneously, amplifying a downturn. Kind of like what happened with margin lending in 2008. Lucas: Yeah, that's a systemic risk that regulators are only beginning to think about. The Financial Stability Oversight Council mentioned it in a report last year — the idea that algorithmic herd behavior in lending could exacerbate economic cycles. Not a lot of concrete action yet, but it's on the radar. Luna: So what do you see as the most promising application of cash flow underwriting that's actually working today? Lucas: I'd say invoice financing for small contractors. A plumber or electrician does a big job, sends an invoice for $10,000, and the client takes 60 days to pay. That plumber can't wait two months. A cash-flow-based lender can look at their history of invoices and payments, see they've always been paid, and advance 80 percent of that invoice within 24 hours. Luna: That's exactly the kind of real-world use case that makes this technology feel less abstract. It's not just about SaaS companies; it's about Main Street businesses keeping their doors open between payments. Lucas: And that's where I think the real transformation happens. Over time, every business will have a dynamic credit line that adjusts to their actual cash flow, not a static limit set once a year. The question is whether we can build those systems in a way that's fair, transparent, and stable. Luna: Speaking of building things — if these conversations are useful for what you're building or running, there is a simple way to help keep the show ad-free and independent. Some listeners have asked about it, so I'll just say: if today's episode was worth a coffee to you, you can find us at buy me a coffee dot com slash fexingo. Lucas: Yeah, exactly. No pressure — just a small way to help if the show is valuable to you. And we really appreciate those who have already done it. Luna: Alright, back to the future of lending. Lucas, you mentioned the static credit limit being a thing of the past — but what about the human element? Are loan officers going away? Lucas: Not entirely. For complex commercial loans — the kind that involve collateral, covenants, or restructuring — you still need judgment. But for the millions of small-dollar loans that businesses need every day, the algorithm can handle it. The human role shifts from gatekeeper to exception handler, focusing on the edge cases the model can't assess. Luna: So the underwriter becomes more like a data scientist. That's a big shift in skills and training. Lucas: It is. And it means community banks, which often pride themselves on relationship lending, will need to invest in technology to compete. Because if a fintech can offer a decision in minutes while the community bank takes two weeks, the relationship only goes so far. Luna: Unless the community bank partners with a fintech to offer the same speed, but with the human touch added. Lucas: That's exactly what some are doing. There's a startup called Amount that helps banks digitize their lending — they plug into the bank's core system and offer a digital application with cash flow analysis. The bank keeps the customer relationship, but the tech enables speed. Luna: So the endgame is a hybrid model: real-time cash flow data powers the initial decision, and human judgment handles the nuance. That feels like a good balance. Lucas: That's the vision. And the data keeps getting richer. Some lenders are now integrating with payroll providers to see not just revenue but also payroll consistency — if a business always pays its employees on time, that's a strong signal of financial health. Luna: That's clever. Payroll is one of the last things a business stops paying when they're in trouble. So seeing payroll continuity is a very positive signal. Lucas: Right. And it's just another data point that feeds the model. The more signals you have, the more accurate your risk assessment, and the more credit you can extend to businesses that deserve it but lack traditional metrics. Luna: It's a fascinating shift from backward-looking to forward-looking finance. I'm curious to see how regulators balance innovation with consumer protection. Lucas: It'll be the defining fintech policy debate of the next few years. And the outcome will shape whether this technology actually closes the credit gap or just creates a faster, more automated version of the same inequalities.