Latest / Fintech Conversations with Fexingo: Banking, Payments, and Financial Software Companies / How Open Banking Data Powers Dynamic Mortgage Underwriting
Transcript
- Lucas: So there's this stat that's been stuck in my head all week: the average mortgage application in the UK still takes over three weeks to process, even though the actual credit decision could be made in minutes if you just looked at the applicant's real bank feed. Luna: Three weeks? I thought we were past that. I opened a bank account in under five minutes the other day. Lucas: Exactly. And that's the gap we're going to dig into today — why mortgages are still stuck in the paper era while every other financial product has gone real-time. The answer, as you might guess, involves open banking and a handful of startups trying to rewrite the underwriting playbook. Lucas: And before we get too deep into the weeds, I want to flag something. This show is ad-free, and it stays that way because listeners occasionally chip in. If today's conversation earns its keep, you know where to find us — buy me a coffee dot com slash fexingo. Luna: It's a small ask, and it genuinely keeps the lights on. Speaking of small asks — how small is the data ask in open banking? That's what I'm curious about. Lucas: Right, well, the whole premise is that you consent to a lender reading your transaction history — typically six to twelve months of current account data. But it's not just a PDF of your statements. It's structured, categorised, machine-readable data that gives a lender a live view of your cash flow. Luna: So instead of trusting your salary slip and a handful of bank statements you print out, they're seeing your actual spending habits. That's a big shift. Lucas: It is. And the interesting bit is what they do with that data. Traditional underwriting looks backward — your credit history, your debt to income ratio. Open banking underwriting looks at your behavior. Are you consistently spending more than you earn? Do you have irregular income? The algorithm can spot patterns a human underwriter would miss. Luna: Like the gig economy worker who has three different income streams and no pay slips. That's the classic example, right? Lucas: Exactly. And that's where I want to take a real case. There's a UK startup called Bud — they've been around for a while, but they've recently pivoted hard into mortgage underwriting. They're not a lender themselves; they sell the analytics engine to banks and building societies. Luna: Oh, I've heard of Bud. They started out as a personal finance app, didn't they? Lucas: They did. And they've repackaged their technology into something they call 'cash flow insights.' The pitch to lenders is simple: instead of asking for three months of payslips, ask for six months of transaction data, and we'll tell you if this person can actually afford the mortgage. Luna: And that's faster, but is it more accurate? I keep coming back to that question. Lucas: The initial data suggests yes, at least for a certain slice of borrowers. Bud ran a pilot with a building society last year — I think it was Yorkshire Building Society — and they found that their model could predict default better than the traditional credit score, particularly for self-employed applicants. Luna: Because the credit score is basically useless for someone with volatile income. It punishes them for not having a steady paycheck. Lucas: Right. And that's the inclusion angle. But there's a flip side, and I want to get to that in a second. First, let's talk about the regulatory tailwind, because that's a big reason this is moving now. Luna: The FCA has been pushing open banking for years, but I feel like the mortgage side is only recently getting attention. Lucas: So, in 2025 the FCA published a report on mortgage market competition, and they explicitly highlighted that lenders aren't using open banking data enough. They argue that better use of data could reduce the number of people who are 'mortgage prisoners' — people who can't switch because they don't fit the traditional lending criteria. Luna: Mortgage prisoners, that's a brutal term. And it's a real problem in the UK — hundreds of thousands of people stuck on expensive deals. Lucas: And open banking won't solve that overnight, but it does chip away at it. Because if you're self-employed and your tax returns are messy, but your bank feed shows a steady stream of income, a lender can now say yes. That's a tangible outcome. Luna: So give me a concrete number. How much faster are decisions actually getting? Lucas: Bud claims they can cut underwriting time from about twenty-one days to under four hours. That's not hyperbole — they've done it in live deployments. The manual verification step is just gone. Luna: Four hours. That's a coffee break, not a waiting period. Lucas: But here's the tension. The same data that makes this possible is incredibly sensitive. You're giving a lender access to every transaction you've made — every takeaway, every night out, every subscription you forgot to cancel. That's a lot of trust. Luna: And it's not just trust in the lender. It's trust in the algorithm. What if the algorithm decides you're risky because you buy cigarettes or lottery tickets? That feels like a slippery slope. Lucas: It is, and there's research on that. A study from the University of Bristol last year found that open banking models can inadvertently penalize people who shop at certain stores or have irregular spending patterns — even when they're perfectly able to repay. It's a classic algorithmic bias problem. Luna: So the promise is inclusion, but the risk is new forms of exclusion. That's the double-edged sword. Lucas: Exactly. And that's why the FCA's involvement matters. They're not just saying 'use more data' — they're also working on a framework for how lenders should use it responsibly. There's a consultation out right now on 'fair value' in mortgage pricing, and open banking data is right in the middle of it. Luna: And I have to ask — what about the customer experience? If I'm applying for a mortgage, I have to connect my bank account, and that's a moment of friction. How are fintechs making that painless? Lucas: That's where the user experience comes in. The good ones use a third-party provider like Plaid or Yapily, and they've gotten really good at it. You get a redirect to your bank's login, you authenticate, and you're back in seconds. The data is pulled in the background, and the lender gets a clean dashboard. Luna: But there's still the 'big brother' feeling. You know, 'you're now sharing your entire financial life with us.' That has to turn some people off. Lucas: It does, and that's why the best fintechs are transparent about it. They show you exactly what data they're pulling and why, and they give you the option to revoke access after the decision. But the reality is, most applicants don't read the consent screens. They just want the mortgage. Luna: Which brings us back to the responsibility on the lender's side. They can't just offload that onto the customer. Lucas: Right. And that's actually where I think the industry is headed — not just using open banking for the initial decision, but for ongoing monitoring. Imagine a lender that checks your cash flow every month and offers you a better rate if your finances improve. That's dynamic pricing, and it's already happening in some credit cards. Luna: So mortgages could become more like a subscription service? That's a big mental shift. Lucas: It is, and it's not without risks. If you have one bad month and the lender reacts, that could be problematic. But the counterargument is that it's better than the current system, where your rate is locked in for two or five years regardless of whether your life gets better or worse. Luna: I want to come back to the bias issue because I think it's the thing that keeps me up at night. How do you test for bias in these models? Lucas: So, the best practice is to run 'fairness audits' before deployment. You feed the model historical data and see if it's treating different demographics differently — not just on gender or ethnicity, but on things like income stability and spending habits. The challenge is that bias can hide in unexpected places. Luna: Like you said, the store you shop at. Lucas: Exactly. And there's a real trade-off between accuracy and fairness. A model that's great at predicting risk might also be great at discriminating. That's the central tension. Luna: So what's the path forward? Is it regulation, or is it self-regulation by the fintechs? Lucas: I think it's both, but regulation has to lead. The FCA is already hinting at requiring lenders to prove their models are fair before they can use them at scale. That would be a game-changer. But it's also going to be slow, and in the meantime, fintechs like Bud are going to keep pushing the envelope. Luna: And I guess that's the story of fintech — the technology races ahead, and the rules chase after it. Lucas: Yeah, that's the cycle. And for now, the tech is winning. Open banking mortgage underwriting is already live in dozens of lenders, and the numbers are growing. I read a report from Open Banking Limited that said usage of their APIs for mortgage applications tripled in the last year — that's a lot of people giving up their bank feeds for a shot at a home. Luna: Tripled. That's not a niche anymore. That's a movement. Lucas: It is, and I think it's a positive one — with cautious optimism. The potential to include more people in homeownership is real, but we have to keep the guardrails up. The question is whether we can build a system that's both fast and fair. Luna: And that's a question we'll be watching. For now, I think we've given our listeners a lot to chew on. Lucas: We have. And if this one stuck with you, you know the drill — buy me a coffee dot com slash fexingo. Thanks for being here.