DEV Community

Swipe Again for Swipe Again

Posted on

The Reply Is the Metric Nobody Puts on the Dunning Dashboard

Open rates get reported because they are easy to count. Click rates get reported because they are slightly harder. But the number that quietly decides whether a dunning sequence works is one many dashboards never surface: how many subscribers wrote back. A reply means a human read the email, understood the situation, and chose to engage with it — and an engaged subscriber whose card failed is usually one clear instruction away from being a retained subscriber.

This is not an argument for folksy copy. It is an argument for writing each email in a sequence as if a person will answer it, because some will, and the answers are diagnostic gold. Replies tell you which instruction confused people, which deadline felt threatening, which link landed in the wrong account state. A sequence with a healthy reply rate is a sequence that is being read as communication. A sequence with silence and unsubscribes is being read as noise, and noise gets filtered — by people first, then by spam systems that learn from people.

The structure that earns replies is unglamorous. One email, one job. The first failure notice states what happened in plain terms: the card on file did not go through, here is the amount, here is what happens next and when. No alarm, no false urgency — the subscriber's access is not ending today, and pretending otherwise spends credibility the later emails in the sequence will need. The mid-sequence message changes one variable: a shorter path to updating the card, or a specific date tied to a pay cycle. The final notice is honest about the ending and leaves the door visibly open behind it.

What kills replies is sameness wearing different subject lines. Four emails that say the same thing at rising volume teach the reader that opening the next one has no value, and readers learn fast. Each message in a working sequence carries information the previous one did not: a new date, a new option, a new and specific consequence. If an email in your sequence cannot answer the question "what does the reader know now that they did not know before?", it is filler, and filler is what trains the silence you are measuring.

Tone deserves one precise note. Subscribers whose payments failed are not delinquent; they are customers mid-accident. The emails that get answered sound like they were written by the business the subscriber chose on purpose — because they were. Threat-adjacent language borrowed from collections does not merely underperform here; it misidentifies the relationship, and subscribers can tell.

Subject lines carry more of this burden than their size suggests. The subject is the one part of the email guaranteed to be read, including by people who never open the message, so it should carry the plain fact — a payment did not go through — in the voice of the business, without countdowns or warnings the body cannot honestly support. A subscriber who only ever reads your subject lines should still come away knowing what happened and roughly when it resolves. Tested against that standard, most subject lines in most sequences fail, and rewriting them is the cheapest experiment in the whole program.

For frameworks organised by sequence position, dunning emails written to be answered, not just opened breaks down what each message in the sequence should carry and what it should leave out. Adapt the structures, keep your own voice, and start counting replies alongside recoveries for one month. The correlation tends to make the argument by itself.

A sequence people answer is a sequence that is still in conversation with your customers. That conversation, more than any single retry, is what recovery is made of.

That is the standard Swipe Again writes to: short, human, one clear action per message, and an ending that respects the reader either way.

Top comments (0)