Subscriptions in crypto — nobody can pull funds, so what is everyone actually doing?

A card subscription works because the merchant initiates the charge. Crypto has no equivalent; the customer initiates every payment, always. That single difference breaks the entire recurring model and I have not found anyone writing about it honestly.

My product is billed monthly. Moving it to crypto means that every month, every customer has to remember to do something. I already know what that does to a renewal rate and it is not a small number.

So what are people running in practice? I am less interested in which gateway has "recurring" in its feature list than in what the actual retention looks like once the customer has to press the button.

there is no subscription. there's a prepaid balance that runs out. sell six months up front and stop pretending

This is a solved problem on-chain and has been for years. The customer signs an allowance to your contract once, you draw against it monthly, they revoke whenever they want. Pull payments exist, they just don't exist inside a hosted gateway's dashboard.

And what share of a normal customer base signs an open-ended token allowance to a company they bought a $19 product from? Post the number if you have it.

What the gateways call "recurring" is an invoice on a schedule plus reminder emails. Plisio and NowPayments both have that in the API. Nothing pulls anything — it generates invoice number 4 on the same day of the month and mails a link.

Which is fine, and it is genuinely most of what you need. The part that decides whether it works isn't the billing, it's the dunning: how many reminders, over how many days, and what happens to access on day 4 versus day 14. That logic is yours to write in both of them, and the gateway will not help you.

Realistically expect a chunk of involuntary churn that a card subscription would never have had. People don't refuse to pay, they just don't get round to it, and then they've stopped.

That matches what I feared. For reference my current voluntary churn is a little over 3% a month; the question is what the involuntary layer adds on top.

IME the grace period is the whole product. Too short and you cut off people who were going to pay on Tuesday, too long and you're giving the thing away.

# what the state machine has to have, whatever generates the invoice
due -> reminder(d+0) -> reminder(d+3) -> soft_lock(d+7) -> reminder(d+10) -> hard_lock(d+21)
# soft_lock = read-only, they can still see their data and the pay button
# hard_lock = the only screen is the pay button

Soft lock is the one that pays for itself. Cutting straight to hard lock on day 7 cost me about a fifth of the recoveries when I tested both.

BTCPayServer does the scheduled-invoice half of this fine if you're self-hosting anyway, but you are still writing all of the above yourself.

My friend, one thing nobody say here — actually monthly is the problem, not crypto.

For my customers in Brazil the fee to move money each month is more annoying than the price of the product. Same the confirmation waiting. When i change to quarterly with small discount, the churn actually go down, not up. Less times to remember, less times to pay fee.

Is not for everybody but if your price is small, monthly is expensive for the customer in a way that don't show in your dashboard.

four reminders and a soft lock. congratulations, you've built a debt collector to chase nineteen dollars. sell the year

Annual works until you want the revenue to be predictable rather than lumpy, and until someone asks for a refund in month two. It's a real option, it isn't a free one.

Reporting back for the record. I have moved to quarterly with a modest discount and an annual option, largely on the strength of the LATAM point above, which I had not considered at all.

Two weeks is not data. What I can say is that the tooling question turned out not to be the question; every gateway in this thread does the same thing and the difference is entirely in what I write around it. Leaving this unresolved rather than picking a winner, because I do not think there is one.