That's a shame, but it seems like it's enabling resellers to charge 25% margins. I don't see those margins being competed out.

Imagine you have services that the end-customer has subscribed for. Some of them are billed weekly, some monthly. Some of the monthly ones are billed the middle of the month based on usage, and some at the week following the month's end based on a mix of monthly subscription and overage.

When a specific invoice that has the potential to vary greatly from period-to-period, you might encounter cases where you have to pay for an invoice that was twice the amount of last month, but you will only be billing your end-customer 20+ days from now! This means that you have to either foot the bill, or wait a little bit and keep the amount in your balance sheet.

When you scale up, you are essentially providing a financing product. You can counter this by asking the users to deposit an amount equaling the base subscription or other service fee, but you can still run into cases where you might simply be cash-strapped to pay your clients' bills; this would put the hypothetical reseller into essentially what's a bank run. Not fun.

In Sweden we also have something similar to purchase orders which is common even with B2C and it always bothered me how often it is used compared to the US where credit card was more normal. I think the article brings up a good point, if a startup could figure out the taxes and bookkeeping, set some reasonable margins or costs for the service it could be a really good business.

Edit: Sorry, they would handle the POs so that is liability, I guess some kind of verification or legal framework is also needed here.

>In Sweden we also have something similar to purchase orders which is common even with B2C and it always bothered me how often it is used compared to the US where credit card was more normal.

Our clients send us purchase orders in Europe, but the context is custom, turn-key, machine learning products developer for large enterprise. We take them by the hand from "I have many problems" to a product their people want to use.

We're building our machine learning platform to make that product development faster (data, scheduling training notebooks, automatic tracking, etc.). We also adopted a plugin architecture for product so we can plug functionality and, more importantly, remove functionality to allow clients to discard what they don't need and avoid a product with 2000 buttons and menu items.