Consumer Credit is Challenging - ProductFTW #96

We think about product in the very broad sense: product managers are responsible for everything in and around their product.

As a team, we help build, launch, and manage five or six credit products each year. Consumer revolving credit is one of the most complex financial technology products.

We’ll dig into each of these more in future posts, but first we wanted to highlight a few of the top reasons why building credit card products is so hard.

1. Underwriting is a multi-step process with many variables

I wrote about underwriting over on CardsFTW (Underwriting is More Than Your Score - CardsFTW #210), so you can check that out for more detail. In short, though, while many people think of underwriting as simply “is my score good enough?” there are many factors. At the start are knock-out rules, such as minimum score thresholds, but there can be knockouts on a dozen or more separate credit factors, such as income, DTI, history length, tradeline counts, delinquency recency, inquiry counts, and utilization. We also have to calculate the right credit APR band, credit limit, the impact of the new credit we offer, and bank-specific preferences.

Plus, people’s credit files are weird, so there are inevitably edge cases and strange results to adjust.

2. The posted APR is not what the business actually earns, and it's not what most cardholders actually pay

Good product managers know how their product generates revenue and profit. It’s easy to think that if you charge 24.99% in your middle band, then you’ll probably make 24.99%. It's never this simple. Subprime cardholders pay interest and fees at higher rates than prime. Weave in introductory balance offers, initial grace periods, payments at various times of the month, and the amount of interest you are earning is usually less than the posted APR. At the same time, the effective cost can be higher for some users.

3. Interchange is not what it seems

The headline rate of interchange bites programs early and often. Major merchants may have special programs (ahem, Amazon, Costco). Your users may love mobile wallets more than others. Your users may travel internationally. Each of these causes your interchange to adjust. You can’t ask your bank or network before you launch either. They’ll basically say: it’s a big, complex table; wait and see.

Key point: assume it’s worse than your model says.

4. The regulatory stack isn't one law; it’s a bunch of them.

Credit programs must keep track of BSA, Reg B, Reg Z, SCRA, MLA, and more. Laws can be unclear, laws can create conflict with each other, and every bank thinks your old bank’s position is absurd.

5. Fair lending complicates even the variables that feel objective

Disparate treatment and disparate impact are two distinct risks under ECOA, and a facially neutral variable (alternative cash-flow data, a platform metric) can still fail if it correlates with a protected class without documented justification. The adverse action requirement compounds this: you have to give specific, accurate reasons for a decline, and per CFPB, a complex or black-box model does not excuse a vague reason. So the more sophisticated your underwriting gets, the harder your compliance obligation gets, not easier.

6. Who is delinquent and by how much really matters

There’s a simple way to look at this:

• Paid on time
• Paid late (1-30)
• Paid very late (31-180)
• Charged off

However, you could also split it into likelihood to go delinquent (always pays the minimum), first time you’re late (what’s wrong with a day or two), and awareness that, in some cases, that first late payment leads to a full charge-off.

Once you’re past 180 days and in collections, you may get some pennies back on the dollar, and it is all a part of the equation.

7. You need a lot of money to lend

Most fintech companies cannot finance the lending of their card program themselves. Think about it this way: 2,000 cardholders spending about 2,000 a month and revolving some of it can require $5-10MM in tied-up capital at a steady run-rate.

As a result, fintech card programs need a warehouse lender for most receivables. Early-stage companies often can't get a second lender for that residual piece, so they have to raise equity to lend, not just to run the business. The private credit funds have their own concerns.

8. The rules can change under you after you've already built for them

In 2024, the CFPB planned to limit late fees to $8. In 2025, the rule was vacated. In 2026, Senate democrats tried to introduce it again. What’s next? Who knows? I do know you’re responsible for it.

9. Fraud and credit risk are separate axes, and fraud is oddly hard to benchmark by tier

Net fraud losses for prime and super-prime run roughly 5 to 12 basis points of purchase volume, with best-in-class programs at 3 to 5 bps, but public data doesn't segment fraud by FICO band. Even a "simple" question like "what's normal fraud loss for a prime card" takes real digging because the industry doesn't publish it that way.

10. Calculating interest is hard

You can’t get this one wrong; it’s against the law to do so. You must calculate it correctly. If a user is revolving but disputes a transaction and then wins, it’s one scenario. They lose; it’s another. They win, but you miscalculated the required payment and charged interest; that’s not good. A lot of companies have recently entered revolving credit and say they know what they are doing, but it’s ultimately a lot more complex than that. You must test and ensure every scenario is run through.

Next time

We’ll dig into each of these in more detail over the coming year. What’s your credit card gotcha you want us to discuss?

About ProductFTW

ProductFTW is a biweekly newsletter about product management, with a focus on real-life experiences in startups. We want to help product leaders succeed by offering realistic approaches that aren’t for giant tech companies. We know you don’t have a full-time product designer on each team. We know your software probably hasn’t been used by millions of people worldwide–yet. We’re here to bridge the content gap from building your product and team to scaling it.