I agree with unmole that it seems AI generated. The original comment (which was later edited to the one we see now) had even more clearly an LLM style to it:

That’s a fair concern — we definitely thought about the ethics side when building this. What we’ve seen in practice is that most “failed payment” churn isn’t intentional churn. Customers still want the service, but their primary card expired, was replaced due to fraud, or hit a temporary limit.

When we tested this, refunds and chargebacks were actually lower for the recovered cohort compared to baseline. That suggests customers were happy the service continued, rather than surprised or upset.

So while it might sound like “rifling through a wallet,” the reality is we’re just using Stripe’s stored, valid payment methods (with the same PCI and authorization framework) that the customer added for this merchant. For customers who really don’t want the subscription anymore, they can still cancel as usual — this doesn’t override cancellations.

That plus the em dashes feels very ChatGPT.

Intuitively you'd think that would be the case but in practice once a subscription stops, a large share of customers don’t come back even if they were active. We see three buckets: 1. Customer shops and finds an alternative (Claude instead of ChatGPT) 2. Customer is pissed off about downtime and leaves 3. Nice to have and customer may reactivate later or must have and reactivates/shops

Ultimately, you want to retain customers that want to use your product. If they don't they will actively cancel. We don't prevent or get involved with customers that actively cancel.

The service didn't stop. You subscribe to a product that you're using but the bank declines the payment (happens every time to almost 30%+ of customers). The customer never intended to churn.. they can always go and cancel. It's good customers with legitimate payments.