Call one. A subscriber's renewal failed on Tuesday — card replaced by her bank, new one in a drawer she hasn't opened. She doesn't know anything failed. She likes the product; she used it yesterday. What she needs is a short, friendly note with a link, and about ninety seconds.
Call two. A business account is ninety days past due on an invoice it has acknowledged twice and paid never. The relationship is over in every way that matters; what remains is a debt, and the work now is documentation, leverage, and eventually a decision about how much recovery justifies how much cost.
Both calls get described, in sloppy operations language, as "chasing money owed." They share almost nothing else, and confusing them is expensive in both directions.
Treat call one like call two and you get the classic dunning disaster: escalation language aimed at someone who never decided anything. Words like "past due," "final notice," and "action required to avoid termination" land on a loyal customer as an accusation. Some pay and quietly downgrade their opinion of you. Some pay and cancel anyway, because who wants to be talked to like that by a service they chose? The money recovered in that moment is real, and so is the trust spent to get it — and trust, unlike the renewal, does not automatically come back next month.
Treat call two like call one and you get the opposite failure: endless courteous nudges sent to a counterparty who has already decided not to pay, while the trail of evidence a real collections process needs goes unbuilt. Politeness has a statute of limitations too.
The cleanest way to keep the two apart is to ask what the other side's state of mind is. In dunning, there is no dispute and usually no decision — the payment failed for a mechanical reason and the customer's intent to stay is intact until you damage it. In collections, intent is precisely the thing in question. Everything downstream follows from that single distinction: tone, cadence, channel, and the point at which you stop.
It is worth reading where dunning ends and collections begins as a boundary you draw before you need it, not during a bad month. Most subscription businesses live almost entirely on the dunning side of that line — involuntary churn is a machinery problem, not a delinquency problem — and the rare account that genuinely crosses into collections territory should cross deliberately, with a handoff, not drift there because a retry sequence ran out of steps and kept escalating by default.
A useful test for any message in your recovery sequence: read it as the subscriber in call one. If it would make her feel accused, it is a collections message wearing a dunning costume, and it will cost you customers whose only crime was a reissued card. Then read your oldest unpaid invoice against the call-two standard: if you are still sending it friendly reminders at day ninety, you do not have a collections process — you have a hope, formatted as an email.
The boundary also protects the metrics. Dunning recovery rates measure machinery: how well retries, card updates, and kind messages rescue good customers from bad luck. Collections recovery measures something else entirely, against older, colder, disputed money. Blend them and both numbers become useless — the dunning rate looks mysteriously terrible, the collections effort looks mysteriously gentle, and nobody can tell which machinery to fix.
Keep the calls separate, and each gets the treatment it deserves: the subscriber gets ninety seconds and a link; the debt gets a file and a decision.
Swipe Again works strictly on the dunning side of that line: pre-dunning where a failure can be seen coming, decline-aware retries aimed at paydays, branded messages that never sound like a collections letter, and recovery counted only on the original failed dollars — flat $179 a month, with a 90-day backfill for the backlog.
Top comments (0)