Per-seat pricing has always felt annoying to me. Ideally you want everyone at your company to be able to access the tools that everyone else uses (you don't want to create a bunch of second class citizens), but that gets really expensive really quickly. So per-user-per-month has never felt great to me.
That said, usage-based billing is confusing. And when customers are confused, they simply chargeback the credit card charge. AWS can make it work, but I'm not sure if everyone else can. Will be interesting to see.
this is not good faith B2B behavior. i'd be surprised if 1% of B2B usage based billing customers behave like this.
It's something you have to expect, and kind of gauge how much money the customer has and send them lots of email as they approach a limit, and make it very clear in the UI that money is being spent that they will soon have to pay. Even then, who knows what people will do. It's bad faith on their part; you, the service provider, are the ones penalized.
I remember having to negotiate with Stripe the removal of their lowest limits. The quantum of unit I was billing on my cloud platform was such that you could consume just enough CPU and storage to end up with a $0.01 invoice at the end of the month.
It was impossible for Stripe to process this payment. IIRC their lowest limit was $0.50 or $0.10. I guess they had this limit in place to prevent abuse and limit fraud. As we had similar hard-coded heuristics for the same reasons.
My most recent Rackspace bill was $0.05. I don't even remember what I'm using it for any more, but it's not worth it to check out.
My last Pulumi bill was $0.86.
That's worth investigating!
Now in a fictional, adversarial way, you can think of a bot exploiting this discrepancy to bankrupt a competitor: create dozens of fake accounts, consume just shy of $0.01 of resources, and have the platform pays 30x more in payment fees (public Stripe price is 30¢/transaction).
To incur a net loss of $1,000,000, you’ll have to find an antagonist ready to create 3,448,276 accounts (1_000_000 / (0.30 - 0.01)), each with their own identity and mean of payment, for a total of $34,482.76.
Fortunately this is highly impractical and will be caught real fast by your internal anti-abuse systems (you have those in place right?).
Small players would be impacted by a 100k$ net loss...and you could even spread it out over several months to be less likely to be caught by anti-abuse systems...
I agree that usage-based billing can be difficult to predict revenue from. We don't think companies will switch entirely to this model. However, the hybrid model can be beneficial for both customers and vendors. Companies can charge a base platform fee and then charge for overage instead of blocking it. This could increase revenue for companies that currently use subscription-based billing.