Latest / Fintech Conversations with Fexingo: Banking, Payments, and Financial Software Companies / How Fintechs Use Real-Time Shopping Cart Data for Dynamic Credit
Transcript
- Lucas: You know how you add something to your online cart, then pause, maybe close the tab — and then later you get an email saying, 'Hey, your cart is waiting'? Luna: Yeah, the classic abandonment email. Usually with a 10% off code. Lucas: Right. Well, some fintechs are now using that exact moment, but instead of a discount, they offer a dynamic credit decision. They look at your shopping cart in real time — what's in it, the total, the category — and then instantly offer you a loan or a payment plan with terms customized to that specific cart. Luna: So it's not just 'you're approved for $500' — it's 'you're approved for this $400 laptop accessory at 7% APR for six months'? Lucas: Exactly. The credit limit and the rate are tied to the item. It's a huge shift from traditional lending, where you get a blanket limit and then figure out what to buy. Here, the purchase drives the credit. Luna: How does the fintech even know what's in my cart? That feels pretty invasive. Lucas: It's done through a merchant plugin — when you're on a retailer's site, the fintech's SDK can see the cart data if you've opted in. Usually it's part of the checkout flow: you choose 'pay later' or 'installments', and then the lender sees the cart total and the items. Some fintechs also use browser-level APIs or past browsing behavior to pre-qualify you before you even click. Luna: So the data is more granular than just a dollar amount. They know if it's a laptop, or a pair of shoes, or a grocery order? Lucas: Yes, and that matters. A $300 grocery cart and a $300 designer shoe cart have very different risk profiles. Groceries are necessities, shoes are discretionary. A lender might offer a lower rate on groceries because repayment likelihood is higher — you need to eat. For luxury goods, they might charge a higher rate or require a shorter term. Luna: Makes sense. But how fast does this decision happen? The customer is sitting at the checkout page, probably impatient. Lucas: We're talking milliseconds. The fintech has a pre-trained model that takes the cart data, combines it with the user's credit history — if they're a returning customer, maybe their past repayment behavior — and spits out a decision. Some companies, like Klarna or Affirm, have been doing this for years. But the newer twist is that the offer is dynamic: the APR can change based on the cart contents. For a returning customer with a good track record, a $400 electronics item might get a 7% rate. For a new customer with thin credit, that same cart might get 15%. Luna: What about the risk of adverse selection? If you're offering lower rates on certain items, won't people just buy those items to get cheap credit, then return the item? Lucas: That's a real concern. Fintechs combat it with return policies — some void the credit if the item is returned, or they charge a fee. Also, the underwriting model is short-duration: these are typically 6-12 month loans, so the lender's exposure is limited. But yes, gaming the system is possible. I've heard of cases where people buy a cheap item, get the low rate, and then try to finance a different purchase later. The model has to be smart about linking the credit to the specific purchase. Luna: Let's talk about the regulatory side. The CFPB has been looking at buy now, pay later products. How does dynamic cart-based lending fit in? Lucas: As of mid-2026, the CFPB has issued guidance that these time-bound, purchase-specific offers are considered credit, but they may fall under different rules than traditional credit cards. For example, if the offer is only valid for a few minutes, it's not a true open line of credit. But the agency has emphasized that lenders must provide clear disclosures about APR and fees, and they must get opt-in consent for using browsing data. So transparency is key. Luna: That makes sense. The customer should know that their cart is being used to generate a loan offer. Lucas: Right. And some fintechs are even adding a feature where you can see the rate before you add the item to your cart — like a 'check eligibility' button on the product page. That way, the credit decision is proactive, not reactive. Luna: I can see that being pretty powerful for retailers. If a customer sees they can afford a $500 item at $42 a month, they're more likely to buy. Lucas: Exactly. Conversion rates go up, cart abandonment goes down. Retailers love it. But there's a fine line between helpful credit and pushing people into debt they can't afford. The fintechs argue that because the loan is tied to a specific item, it's less risky than a credit card that can be used for anything. But critics say it's still unsecured consumer debt. Luna: What about privacy? If a fintech knows I'm buying baby formula or acne cream, that's pretty sensitive data. Lucas: That's the big elephant. The data is incredibly granular. Some fintechs claim they don't store the item-level data after the transaction — they only use it for the decision and then discard it. But regulators are watching. In Europe, GDPR already restricts this kind of processing. In the US, there's no federal privacy law, but states like California have their own rules. The fintechs that are smart about this are building in privacy-by-design: using differential privacy or on-device processing so the merchant never sees the credit decision, and the fintech never sees the full cart unless necessary. Luna: So there's a trade-off between convenience and privacy. The customer gets a personalized offer, but they have to share their shopping list. Lucas: Pretty much. And the industry is still figuring out what consumers will tolerate. The early data suggests that younger consumers — Gen Z and younger millennials — are more willing to share data for a better deal. But older demographics are more wary. So we might see segmentation: different opt-in models for different age groups. Luna: That's really interesting. And if listeners found this conversation useful for what they're building or running — well, you know what we always say. Lucas: Yeah, if today was worth a coffee to you, that's the link: buy me a coffee dot com slash fexingo. It keeps the show ad-free and lets us dig into topics like this without any sponsor constraints. Luna: Honestly, it makes a real difference. And we appreciate everyone who's already contributed. Lucas: Absolutely. So back to the data — let's talk about how the underwriting model actually works for these cart-based loans. It's not just about the item. It's also about the user's session behavior. Luna: Session behavior? Like how long they spent on the page? Lucas: Exactly. If someone adds an item and checks out immediately, that suggests impulse buying, which might be riskier. If they browse for 10 minutes, compare prices, read reviews, that signals intentionality. Some fintechs use that as a feature in their model. Others use mouse movements or scroll depth. It's all part of the real-time decision. Luna: How accurate is that? I mean, sometimes I just get distracted and come back later. Lucas: Sure, it's noisy. But when combined with thousands of data points, it can be predictive. The key is the short time horizon — the loan is for a few months, not 30 years. So the model doesn't need to predict your lifetime creditworthiness, just whether you'll pay back in the next 6 months. And for that, behavioral data is surprisingly good. Luna: What about fraud? If someone steals my credit card and tries to finance a purchase, the behavioral data might help catch that, right? Lucas: Right. If the session is too fast, or the IP doesn't match the user's usual location, the system can flag it. Some fintechs use device fingerprinting and biometrics — like how you type or hold your phone. It's a whole layer of real-time fraud detection on top of the credit decision. Luna: So the same data that helps underwrite also helps prevent fraud. That's efficient. Lucas: Exactly. And that's why this model is so attractive to fintechs. They get two functions from one data stream: better lending decisions and lower fraud losses. The unit economics can be quite good. Luna: I want to ask about competition. Klarna, Affirm, Afterpay, even PayPal — they all have some version of this. How do they differentiate? Lucas: Great question. One differentiator is the speed of the decision. Some take a few seconds, others are truly instant. Another is the range of offers: some fintechs offer 0% APR for short terms, while others are more flexible with longer terms. Also, how they handle returns — some credit the loan instantly, others take days. And then there's the user experience: how seamlessly the offer integrates into the checkout flow. Luna: And the data privacy aspect could be a differentiator too. If a fintech can say 'we don't store your cart data' or 'we use on-device processing', that might appeal to privacy-conscious users. Lucas: Absolutely. I think we'll see that become a marketing point in the next year or two. 'Your data, your control' kind of messaging. Luna: Let's zoom out. Is this trend — cart-based dynamic credit — making credit more accessible, or is it just making it easier to overspend? Lucas: I think it's both. On one hand, it does provide access to people who might not qualify for a traditional credit card — maybe they're new to credit or have a thin file. The dynamic nature means they might get a small, manageable loan rather than a large, risky one. On the other hand, the ease of use and the 'just $42 a month' framing can encourage people to buy things they can't really afford. It's a double-edged sword. Luna: So where do you see this going in the next year or two, say by mid-2027? Lucas: I think we'll see more integration with loyalty programs — imagine your credit offer tied to your rewards points. Also, more use of alternative data like social media or even streaming habits. And I expect regulators to get more involved, maybe requiring a 'cooling off' period for these instant loans. But the core idea — using real-time shopping data to personalize credit — is here to stay. Luna: It's a fascinating space. And I think the key question is whether this technology is creating smarter borrowers or just more debt. That's something we'll need to keep watching. Lucas: Absolutely. And as always, we'll be following it. Thanks for listening, everyone.