While reading today’s Stratechery update on Patreon changing their pricing model I came across the following passage:

Of course this wasn’t great for creators: payments made for a creation released on November 2 wouldn’t land in creators’ bank accounts until some time after December 1; the reasoning, though, was clear — credit card fees. Here I can certainly bring personal experience to bear: credit card fees are really expensive (they are by far Stratechery’s largest expense)! While the exact amount varies by network (that is, American Express is the most expensive), most credit card fees are ~$0.30/charge plus anywhere from 1.5% to 3.5% of the total amount charged.

That $0.30 is more important than it might seem; to use Stratechery as an example, I charge $10/month or $100/year — that suggests that I earn $20 less from annual subscribers than I do from monthly ones. However, because I have to incur that fee 12 times a year for monthly subscribers, but only once for annual subscribers, the difference is actually only about $16 (by the way, monthly subscribers can easily upgrade to annual accounts; I push a seemingly lower revenue option because it turns out annual subscribers have significantly higher lifetime value — a higher up-front commitment ends up being a longer commitment). The importance of that fee looms even larger for Patreon’s “charge-per-creation” creators, who often charged only a dollar or two for their creations: were Patreon to charge credit cards immediately, over 30% of payments would go towards fees; however, by bundling payments over the entire month fees are much more manageable.

Clearly credit card issuers and banks need to make some money on their credit cards but these rates seem a bit aggressive, I don’t know the details but I’m sure a chunk of that is to cover cases of fraud. That makes sense but it seems there’s an opportunity to charge lower and lower rates for each recurring subscription charge. The intuition is that the likelihood of a person not being able to cover the subscription charge is much higher the first time they get billed rather than each subsequent one since they’ve built a strong payment history. Beyond this, it does feel as if there’s something a bit more intelligent that can be done around these merchant rates. They were instituted decades ago before the growth of modern data analysis methods that can be run for each individual and had to be set for the whole population. Imagine how optimized these rates can be now given modern tools. This is a huge opportunity and I’m surprised there haven’t been any companies trying to tackle it - maybe it’s due to the complexity of the space or network effects but it just feels as if there’s something here.