I think the answer is more about culture than moats. The big issuers all beefed up their analytics departments when Cap One started to be successful -- often by hiring Capital One staff. The difference was that they had mass market opportunities whereas C1 was had an outsized focus on Subprime. They did other cards, particularly affluent, but they were the biggest issuer focused on Subprime/Near Prime. The other big issuers weren't willing to price-for-risk for reputational reasons, they wanted to minimize risk. So their analytics teams would marginally improve approvals at the lower end, but still wouldn't touch the higher risk stuff. For example, when I was at JPM, we had an APR cap at 29.9%, so we just turned down anyone who wouldn't qualify at that cap. I think Cap One was willing to price above that level when warranted and or layer on additional fees to cover the extra risk.
Very interesting as always! One follow-up: Was Capital One potentially more successful because deep risk-based analytics / credit modeling was inherently harder to replicate, compared to other risk-based innovations (e.g., merchant cash advance, BNPL), which appear to rely more on data advantages or redesigned product mechanics that (within the context of risk) seem to be more easily competed away/copied?
I also wonder whether AI/ML will serve to accelerate the commoditization of such risk analytics / credit modeling strategies as well, making it even harder to differentiate on risk in the future.
Andrew, your breakdown of risk-centred payment strategies is insightful - especially the idea of using analytics to uncover overlooked opportunities. Exploring practical approaches to risk and working capital can add another layer to these strategies. TCLM provides resources that could be helpful in this context.
I think the answer is more about culture than moats. The big issuers all beefed up their analytics departments when Cap One started to be successful -- often by hiring Capital One staff. The difference was that they had mass market opportunities whereas C1 was had an outsized focus on Subprime. They did other cards, particularly affluent, but they were the biggest issuer focused on Subprime/Near Prime. The other big issuers weren't willing to price-for-risk for reputational reasons, they wanted to minimize risk. So their analytics teams would marginally improve approvals at the lower end, but still wouldn't touch the higher risk stuff. For example, when I was at JPM, we had an APR cap at 29.9%, so we just turned down anyone who wouldn't qualify at that cap. I think Cap One was willing to price above that level when warranted and or layer on additional fees to cover the extra risk.
Very interesting as always! One follow-up: Was Capital One potentially more successful because deep risk-based analytics / credit modeling was inherently harder to replicate, compared to other risk-based innovations (e.g., merchant cash advance, BNPL), which appear to rely more on data advantages or redesigned product mechanics that (within the context of risk) seem to be more easily competed away/copied?
I also wonder whether AI/ML will serve to accelerate the commoditization of such risk analytics / credit modeling strategies as well, making it even harder to differentiate on risk in the future.
Andrew, your breakdown of risk-centred payment strategies is insightful - especially the idea of using analytics to uncover overlooked opportunities. Exploring practical approaches to risk and working capital can add another layer to these strategies. TCLM provides resources that could be helpful in this context.
(It’s free)- https://tradecredit.substack.com/subscribe