This should be at the top of the comments IMO. I'm honestly stunned by this blog post because I always assumed a relationship with a payment processor like Stripe was akin to a banking relationship where you'd have an account manager that would reach out to you to resolve problems. If the banks can do it, why can't Stripe? Is it simply a difference in regulation and what they can get away with legally?
All of the big tech companies think they can use machine learning and algorithms to do everything and they have an "acceptable" rate of failure as a target.
The main problem with that is that even if the failure rate is .01%, the failure is typically catastrophic for that .01%. When the error is going to ruin someone's life, is there really an "acceptable" rate of failure?
A secondary problem is that machine learning and algorithms are going to have a tough time accounting for virility. IE: If I have a small product that goes viral, as a percentage change, my error/fraud/dispute rates are going to jump drastically. So at the exact moment where reliable, scalable payment processing is the most important in my life, the automated systems are going to have the highest risk of banning me and automatically denying my appeal.
The fact that 24-48 hours is considered an acceptable timeframe for an appeal is worthy of it's own paragraph. That's unacceptably slow if they're locking the account and doing irreparable harm to your business. That wouldn't be tolerated in a market with proper competition and my instinct is to ask for regulation that would involve a 3rd party in dispute resolution for a payment processor that's terminating a relationship in a non-amicable manner.
At least give me some options that can make things suck less. I'd prepay $500 (non-refundable) without even thinking about it to be guaranteed a phone call prior to account termination. I'd let them hold back a percentage of revenue up to an absolute value so it can be held as a (refundable) bond to protect against fraud. I'd let them hold back a higher percentage if their automated systems detect an increased chance of fraud / issues.
I think stuff like this is a stunning failure and I can't understand how tech entrepreneurs (of all people) can't understand why it's unacceptable. The dream for most of us is literally to build something that has overnight, viral success and makes us rich, but we've got companies like Stripe using ML algorithms that'll auto-ban you as soon as you deviate from the norm. How is that reasonable?
The absolute worst case scenario for a Stripe customer should be for the customer to opt to have all payments withheld (by Stripe) and to undergo some kind of dispute resolution or problem solving. Would you rather wake up to a banned account or an email saying they're holding your money until you call them? I know PayPal gets a lot of flack for the latter, but maybe it's not that bad compared to the alternative. The problem with PayPal AFAIK is they hold the money for a long time no matter what.
I get so frustrated when I see PR / damage control and the solution they're providing is "we're going to improve the algorithms." You can't. By the time those systems fail you need one-on-one human support where both sides can adapt, compromise, negotiate, etc. in real-time.
YOU NEED PEOPLE, NOT MACHINES!