Dunning Email Sequences

A fifth to a third of your churn has nothing to do with your product. Somebody’s card expired, a bank flagged a recurring charge from a US merchant as unusual, a finance team replaced the corporate card after a fraud alert and forgot the twelve subscriptions attached to it. The customer still wants the software. They just stopped paying for it by accident, and in most companies no single person is accountable for getting them back.

How much of your churn is actually a failed card

Pull your last six months of cancellations and split them by cause. Involuntary churn, meaning subscriptions that ended on a payment failure rather than a cancel click, typically runs at 20 to 40 percent of total SaaS churn, and it skews higher for self serve products billed monthly on consumer cards.

That split changes what the number means. A product losing 4 percent of logos a month where 30 percent of that is involuntary is losing 1.2 points a month to payment plumbing. Recover half of it and you have added 0.6 points of retention without changing the product, the pricing or the positioning. No acquisition channel offers that arithmetic.

The reason it stays broken is ownership. Billing sits with engineering or finance, retention sits with customer success, email sits with marketing, and a failed card falls through all three. Marketing looks at it and sees a transactional message that is not their job. Finance looks at it and sees a collections problem. Nobody writes the copy. If you want the definition and the wider context, the dunning email entry covers the term; this page is the build.

The retry and email schedule from day 0 to day 14

Six touches over fourteen days, with the retry and the email doing different jobs. The retry attempts to take the money. The email attempts to get the card fixed. Sending an email without a retry behind it, or retrying without telling anyone, both waste the window.

Two decisions inside that schedule matter more than the copy. The first is retry spacing. Attempting the same card three times in one day burns network attempts and recovers almost nothing, because the condition that caused the decline has not changed. Spreading attempts across a fortnight catches payday cycles, corporate card reissues and bank fraud holds that clear on their own. Visa’s acquirer rules cap retries on a single declined transaction at 15 within 30 days and charge fees past that, so the constraint is real even before you consider the customer experience.

The second is decline code branching, which most teams skip.

Sending a soft decline message to somebody whose card was reported stolen wastes a week of the window. Branch on the code your processor returns. Stripe, Recurly and Chargebee all expose it.

Copy rules that keep dunning out of the promotions tab

Dunning email has one formatting job: look like a receipt, not a campaign. Every visual flourish that makes a marketing email perform pushes a payment notice toward the promotions tab or the spam folder, which is the one place it cannot afford to land.

• Plain text or near plain text. One link, no hero image, no logo header, no footer with social icons.
• Send from the product or billing subdomain, not the marketing one. A payment notice that inherits a marketing domain’s reputation inherits its filtering too.
• Send from a person or a functional address a human recognises, like billing@ or a named account owner, never from a no reply address on a payment failure.
• Subject lines that state the fact. “Payment failed for your Growth plan” beats anything with an emoji, a discount or a countdown.
• No unsubscribe suppression. Dunning is transactional, so a marketing opt out must not stop it, and neither should quiet hours or frequency caps.
• One call to action per message, pointing at a hosted card update page rather than a login wall.

That last point is where most sequences quietly fail. If the update card link drops the customer on a login screen, and the cardholder in the finance team does not have a login, the recovery is dead. Stripe, Recurly and Chargebee all issue signed hosted update links that work without an account session. Use them.

Tone matters less than most people think, with one exception: the first message should carry no blame at all. The overwhelming majority of first declines are bank side and temporary. Treating a routine insufficient funds decline as a delinquency notice offends the customer who was always going to pay on day 3 anyway. Escalate at day 7, not at day 0. The same restraint applies across the wider programme, which is why the tonal shift here sits at the opposite end of the scale from the persuasion work in the trial expiry email sequence.

Pairing the email with an in app banner and an admin notification

Email alone recovers a fraction of what it should, because the person who receives billing email is often not the person using the product. In a 60 seat account the card belongs to a finance manager who opens your product twice a year. The daily users have no idea anything is wrong until the day access changes.

The high value threshold is the part teams resist and the part that pays. A $49 a month account gets six automated emails. A $4,000 a month account gets a phone call on day 3, because the difference between recovering it and not is worth more than a week of the marketer’s salary.

Grace period, downgrade and data retention set the tone

Write the policy before the copy, because every sentence in the sequence depends on what actually happens on day 14. The three decisions are how long full access continues, what state the account moves into, and how long data survives.

The position worth taking: read only beats lockout for almost every B2B product. A locked out customer has no reason to come back and every reason to start evaluating a competitor. A read only customer can still see their data, still gets value from looking at it, and restores with one payment. The exception is products where read only access is itself the product, such as reporting and analytics tools, where the downgrade has to remove something real.

Say the retention window explicitly in the day 10 and day 14 emails. “Your data is kept for 60 days” removes the panic that makes people either churn angrily or flood support. It also removes the false urgency that tempts teams into writing dunning copy that sounds like a threat.

Tooling: what each platform actually automates

Most teams start with what their processor already includes and add a layer only when the recovery number justifies it.

Turn on the card account updater first. Visa Account Updater and the Mastercard equivalent refresh stored credentials when a card is reissued, which prevents a meaningful share of expired card declines from happening at all. It is usually a toggle in your processor and it is the highest return five minutes in this entire guide.

What recovery rate to expect and how to check your baseline

Published recovery rates run from roughly 30 to 70 percent of failed invoices, and every vendor quoting a number at the top of that range is quoting their best cohort. The spread is driven by card mix, not by copy. Consumer cards on monthly plans recover best. Corporate cards, prepaid cards and cross border transactions recover worst, sometimes dramatically so.

Measure before you optimise. For one month, log every failed invoice, the decline code, whether any payment succeeded within 30 days, and whether the account was still active at 60 days. That gives you a baseline recovery rate and, more usefully, a decline code distribution that tells you which branch of the sequence to build first. A base that is 60 percent expired cards needs the account updater and a different email. A base that is 60 percent insufficient funds needs better retry spacing.

Then report it in revenue. Recovered monthly recurring revenue and recovered logos are the two numbers that get this work funded. Open rates on dunning email are high and meaningless, since a payment failure notice is opened by almost everyone who receives it. If you need to size the opportunity before building, the SaaS email revenue calculator will put a range on it, and the lifecycle email audit checklist will tell you what else in the programme is missing while you are in there.

Why dunning must never live on the marketing subdomain

Here is the position, stated plainly. Dunning is transactional email about a contract, and it must be sent from the product or billing subdomain, exempt from marketing unsubscribe suppression, and excluded from every frequency cap and quiet hours rule in your platform.

The failure mode is specific and common. A customer unsubscribes from the product newsletter in March. In July their card expires. The dunning sequence runs through the same marketing ESP, respects the global suppression list, and sends nothing. The account cancels in August and appears in the churn report as a voluntary loss. Nobody ever finds it, because the missing sends leave no trace.

There is a second, slower version of the same problem. Marketing sends a poorly targeted campaign, complaint rates rise on the shared sending domain, and filtering tightens on everything from that domain including password resets and payment notices. Splitting sending domains before there is a problem is cheap; recovering a reputation afterwards costs weeks. That whole architecture is worth doing properly, and once it is in place, the same logic governs churn prevention email campaigns and the onboarding flow in the SaaS onboarding email sequence.

One nuance. Being transactional does not license you to add marketing to it. A payment failure email with an upgrade offer or a webinar link at the bottom is a marketing email wearing a costume, and in most jurisdictions it loses its transactional exemption. Keep it clean.

Where dunning still loses

Three honest limits. Dunning cannot recover an account that genuinely wanted to leave and let the card fail as a passive cancellation, and somewhere between a tenth and a quarter of involuntary churn is exactly that. Treating those accounts to six emails and a phone call annoys them and inflates your apparent failure rate.

It also cannot fix a bad billing experience. If your invoices arrive with no line items, your currency is wrong for the customer’s region, or your card form rejects valid European cards because it never implemented 3D Secure properly, the sequence is patching a hole the product made. Cross border decline rates are frequently a product problem misdiagnosed as a dunning problem.

And it degrades. A sequence written in 2024 with a lockout date that no longer matches the billing code, a card update link pointing at a deprecated page, or a decline branch that never fires because the processor changed its codes. Put dunning on the same quarterly review cycle as the rest of the lifecycle programme in the SaaS email marketing hub, and test the card update link every release.

Build it this week

Start with the single email you do not have. Most companies have one automated payment failure notice from their processor, no branching, no in app surface and no retry strategy beyond the default.

Do this before the newsletter, before the nurture track, before anything with a content calendar attached to it. A payment recovery email is the highest revenue per send message a SaaS company will ever write, and the same argument applies to the upgrade path in freemium to paid upgrade emails and to the structured copy you can lift from the SaaS onboarding email templates.