Content Library

Framing fintech: chatting with fintech founder & investor, Aaron Frank, pt 2

Aaron Frank shares his tips for launching a credit card product

P2 framing fintech aaron frank


In this blog series, Framing Fintech, our GM of Fintech and former Head of Growth at Ramp, Charley Ma, catches up with leaders in the industry to chat through lessons learned, fintech trends, and hot takes.

In the new edition of Framing Fintech, Charley Ma sat down with Aaron Frank to get his advice on how new credit card startups should approach launching a credit product.

Aaron Frank is an advisor, consultant, and investor to many fast-growing fintech companies and a longtime friend of Alloy. He runs a project called Not a Fintech Company, which is aimed at democratizing information in the financial services industry to enable people to build and run fintech companies more efficiently. Previously, Aaron was the co-founder and CEO of the credit card startup Final, which was acquired by Goldman Sachs. Aaron stayed on with Goldman Sachs to help them launch the Apple Card.

Part one of their conversation was so jam-packed that we broke it up into two installments. Check out the rest of their conversation below.

At a glance

The key lifecycle events new credit card companies should be thinking about

  • How to think about the full customer lifecycle for a credit card company: acquisition, activation, usage, and retention

  • Why fintech has product market fit, but not product marketing fit

  • How to think about driving retention and activation

Areas where people get things wrong, make false assumptions, or common traps

  • Throughput is often a limiter on being about to hit month-over-month growth goals

  • The paramount importance of having the right onboarding flow

  • What your north star KPI should be

Types of fraud that people should be paying attention to when launching and running a credit card company

  • Fraud mitigation measures will, over time, be driven by your product construct

  • Why catching fraud at origination is critical

  • Monitoring for account takeovers


So someone is convinced they have this unique, new financial product. What are the key lifecycle events they should be thinking about? What are some things to think about that simplify the complexity of running and starting a credit card business?


Every single startup is going to have similar lifecycle events: acquisition, activation, usage, and retention. But acquisition is so much more painful in fintech because it’s so regulated and it’s so much more costly to do everything. How do you get them through the acquisition funnel? For one, what’s your cost to actually take this customer from a marketing inbound lead all the way through ‘accepted?’ And what’s the percentage that is accepted, and what’s the cost per application, and then the cost per accepted application? How many of those do you want to do in a month? How many people do you need to come through top-of-funnel in order to get that growth rate this month?

The nice thing about fintech is that it has product market fit because the market is so massive. What it doesn’t have, in my opinion, is product marketing fit. It’s hard to know how to acquire your customer the right way that gets them onboarded efficiently so that you’re not so deep in the hole that you’ll never make money on them. This is one of the current fundamental problems that people are facing behind the scenes right now. For credit cards, specifically, you don’t know if that person that you spent money on getting through the onboarding process is going to create a debt receivable for you. They might just become a pure transactor, or worse, won’t use the product, in which case you’ll have thin margins on the interchange.

First, how do you market to this customer and then acquire them? How do you get them through the onboarding funnel? I mean, this is what Alloy does — allows you to do testing, run models, and use alternative data sources to get the identity piece done. Since the origination funnel is so high in friction, you also need to optimize at each of the lifecycle stages for continued usage and retention.

And then, it's time to ship them a card and get them activated and using the card. I just spent, say, $50+ just getting them through this funnel; how do I actually get them to use the card? So it's not just a wasted effort.

And then, it comes down to retention. Can you do things that make sure your customers don't start moving spend elsewhere? That gets tricky because it's a cutthroat market. If you think about all the different players, if I'm doing a card that helps someone go from zero-to-one in credit, well, they go from thin-file/no-file, and after six months, they’ve built a credit file. So a larger bank work with the bureaus and say, “Give me all the new credit files so I can advertise to people that used to be no-file/thin-file but are now showing good credit, and offer them my intro card.” So as a new company offering a credit product, you're kinda constantly fighting uphill on all of these things.


You touched on this a little bit, but what are areas that people get wrong? What are the false assumptions or common traps that you warn people to look out for?


Well, for one, I don't think most people are set up for the throughput that they need to hit the month-over-month growth that they’re looking for. It's just incredibly hard from an operational perspective. Everyone almost always underperforms in that to begin with, and so they end up always having to scramble.

You may build the best product, but if you can’t figure out how to get people through an origination flow, it doesn’t matter. You’re never going to be venture fundable because you’re not growing. Something like Alloy gives you workflow management. So it's actually less of a lift. Fewer people can operate more accounts, and you all have performed on this incredibly well over the years.

And then, and the other thing I tell people is to figure out what the right KPI is. If you’re a lending company, it's probably going to be $1 billion in origination until you’re not subscale.


Wow, that’s interesting. I think you touched upon fraud a little already, but when someone's building a credit card, what are the unique areas of fraud that, particularly when someone is starting, you typically recommend keeping an eye out for or thinking about?


So it depends on the product construct. Some companies always physically ship a card — so they don’t have to worry about this — but we gave instant access to credit when you got approved, so if a customer applied between 9:00 pm and 6:00 am, we just made sure to flag it. Essentially, when the US is asleep and not really applying for credit, should you be approving credit with instant access, or fallback to industry norms and ship them a card?

Over time we got better at that, and it wasn’t always flagged, but it saved a lot of money over the short term while we figured out other signals. If someone complains that they don’t have access to their card when you haven’t even approved them yet, it’s usually a sign of something weird anyways. Obviously, with the US laws and regulations, you're trying to approve as many people in a fair way, but you're also trying to keep out fraud to run a sustainable business.

The other thing I tell people to lean into heavily is to stop fraud at the source. Start at origination. Stopping fraud at transaction time on new accounts is infinitely harder. However, there are things like velocity controls that you can also have. Really think of this as an onion. You're going to have to keep on adding layers and building risk scores at almost every single consumer interaction point. So origination is the big one because you get to stop it upfront.

Once we're into the system, you're going to look for account takeover (ATO) issues and transaction issues, and all these things. But that's more nuanced and hard to pull off. Once you have more data, that also makes it easier. But, that's a problem even Google, Facebook, and Apple struggle with.

Launch your credit card with Alloy

Related content