Latest / Fintech Conversations with Fexingo: Banking, Payments, and Financial Software Companies / How Fintechs Are Using Payment Data for SMB Credit Scoring
Transcript
- Lucas: So there's this stat I keep coming back to — roughly 70 percent of small businesses that apply for a traditional bank loan get turned down. Not because they can't pay it back, but because the bank's model can't see their revenue clearly enough. Luna: It's the classic chicken and egg problem. You need credit history to build credit history. But a lot of SMBs run mostly on cash or debit, or they use multiple payment platforms that don't talk to each other. Lucas: Exactly. And that's where the fintech play gets interesting. Instead of waiting for a credit bureau report, a growing number of lenders are using payment processing data directly — the actual dollars flowing through a merchant's point of sale or online checkout. The idea is: if you can see daily revenue, you can lend against it in near real time. Luna: So they're basically skipping the credit score and going straight to the cash flow. Who's doing this well right now? Lucas: There's a company called Stitch — based in South Africa, but their model is spreading globally. They built an API that lets lenders connect directly to a merchant's payment processor — think Stripe, Square, or local equivalents — and pull transaction-level data with the business owner's permission. The lender sees average daily sales, seasonality, even customer concentration — like, does 80 percent of revenue come from one client? Luna: That customer concentration piece is huge. A bank wouldn't know that from a credit application. But if you're a wholesaler that relies on one big retailer, a default risk looks very different. Lucas: Right. And Stitch isn't alone. Square itself has been doing this for years with Square Capital — they lend against a merchant's own transaction history on the platform. But what's new is that this is becoming a third-party service. A lender doesn't have to be a payment processor to access payment data anymore. They can use an API aggregator to look across multiple processors. Luna: So the fintech lender gets a much richer picture. But isn't there a risk that the merchant switches processors or has multiple accounts? The data could get fragmented. Lucas: That's the challenge. Stitch's API tries to handle that by linking to multiple processors — they currently support about 15 in Africa alone. But the ideal scenario is when the merchant uses a single primary processor, or when the loan product itself is tied to that processor, like Square Capital. Then the data is clean and the repayment can be automatic — the loan is paid back as a percentage of daily sales. Luna: That repayment structure is really elegant. It reduces the risk for both sides — if the merchant has a slow day, they pay less. It's almost like revenue-based financing for SMBs. Lucas: Exactly. And that's been the model for companies like Shopify Capital and Amazon Lending too. But what's interesting in 2026 is how this is expanding beyond the platform lenders. Independent fintechs like Stitch, or even Stripe with Stripe Capital, are opening up the data layer so that any qualified lender — bank or fintech — can underwrite using payment data. Luna: Let's talk numbers. How big is this market? I know the SMB lending gap globally is estimated around $1.5 trillion — that's the unmet demand. Lucas: Right. And most of that gap is in businesses with annual revenue under $1 million. The traditional cost of underwriting a $50,000 loan is about the same as a $500,000 loan — so banks just don't bother. But if you can automate underwriting using payment data, the marginal cost drops dramatically. Stitch claims their API reduces underwriting time from weeks to minutes, and their lenders see default rates comparable to — or better than — traditional fico based lending for this segment. Luna: I want to push back a little though. Payment data is backward-looking. It tells you what happened, not necessarily what will happen. A restaurant that had great sales last year might be struggling now because of a new competitor across the street. Lucas: That's a fair point. And the lenders I've talked to acknowledge that. Payment data is part of a mosaic — they layer it with other signals like bank account data, business age, and sometimes even social media or reviews. But the advantage is recency. A traditional credit report might be six months old. Payment data is yesterday. If a business is declining, you can spot it in two weeks rather than two quarters. Luna: So it's about velocity of information. That makes sense. But what about fraud? A merchant could inflate their revenue by running fake transactions through their own processor. Lucas: It happens. But lenders have gotten smart about it. They look for patterns — consistent daily volumes, a mix of transaction sizes, and they cross-reference with bank account deposits. Stitch, for example, also offers an account verification API that checks if the payment processor payouts match the deposits in the merchant's bank account. If those don't line up, that's a red flag. Luna: That's clever. So the data can actually self-validate to some extent. How are traditional banks reacting to this? Are they partnering or competing? Lucas: Both. Some banks are building their own payment-data lending products — JPMorgan has been piloting a program with their merchant services clients. Others are partnering with fintechs. But there's also regulatory pushback. In the US, the Consumer Financial Protection Bureau has been looking at data aggregation standards, and there's debate about whether payment processors should be allowed to share merchant data without explicit opt-in. Luna: That's a real tension. On one hand, more data could unlock credit for millions of businesses. On the other, merchants might not realize how much information they're handing over when they click 'accept' on a loan offer. Lucas: Yeah, the consent framework is still evolving. Most api driven lenders use a process called 'recurring data access' — the merchant gives permission once, and the lender can pull data periodically. But the merchant can revoke it. The question is whether that revocation is practical or buried in fine print. Luna: You know, speaking of transparency — and this is a bit of a sidebar — but it connects to something I think about a lot with this show. We cover these really specific tools and companies, and a handful of our listeners have mentioned they actually use some of these services. A few even said they chip in to keep the show ad-free through buy me a coffee dot com slash fexingo. Lucas: It's true. And that support is what lets us dig into niche topics like payment data underwriting without worrying about advertisers. We're not trying to sell you a loan — we're trying to explain how the system works. Luna: Exactly. So if you find these conversations useful for what you're building or running, that's the reason we keep doing them. Back to the data side — Lucas, you mentioned Stitch is South African. Are there regional differences in how payment data lending is taking off? Lucas: Big differences. In Africa, payment data lending is leapfrogging traditional credit altogether because so many businesses don't have formal credit files. Stitch's CEO told me that in Kenya, a lot of merchants use mobile money — M-Pesa — as their primary payment method. So lenders are now pulling M-Pesa transaction data to underwrite loans. That's completely bypassing the banking system. Luna: That's incredible. Meanwhile, in Europe, open banking regulations already require banks to share transaction data — so payment data lending is almost an extension of that. And in the US, it's more fragmented because there's no single regulatory mandate. Lucas: Right. The US has the most potential volume, but also the most friction. There are dozens of payment processors, each with their own data format. That's where companies like Stitch or Plaid — though Plaid is more focused on bank data — come in as the normalization layer. If they can standardize the data, the lending use case scales fast. Luna: Let's zoom out. If payment data lending becomes the norm, what breaks? I'm thinking about credit bureaus — do they become less relevant for SMB lending? Lucas: I think they still have a role, especially for larger loans or longer-term credit. But for short-term working capital — the kind of loan a retailer needs to buy inventory for the holiday season — payment data is faster and more accurate. Bureaus are good at predicting default over 10 years. But for a 90-day loan? You want to know how much they sold last week. Luna: So it's not either-or, it's about matching the data source to the loan product. That makes sense. What's the one metric you think listeners should watch to gauge whether this trend is really taking hold? Lucas: I'd say watch the number of API calls to payment data aggregators. Stitch, for instance, reported a 300 percent increase in API requests from lenders in the first quarter of 2026 compared to a year ago. That's a leading indicator. Also, look at how many traditional banks start offering revenue-based repayment loans — when Chase or Bank of America launch that product, you know it's mainstream. Luna: And I'd add: pay attention to the regulatory decisions. If the CFPB or equivalent bodies create a standard for payment data sharing with clear consent rules, that will accelerate adoption. If they crack down on data use without strong opt-in, it could slow things down. Lucas: Right. So the next 12 months are pivotal. We've got the technology, we've got the demand — the question is whether we can build the trust framework to go with it. Luna: And whether small business owners will feel empowered by the data, not exploited by it. That's the real line to walk. Lucas: Agreed. And that's probably a topic for another episode entirely.