Key Takeaways

• Notices and retry attempts run on separate clocks. Chargebee states that dunning "emails can be sent irrespective of the dunning attempts", so a customer's notice needn't match the last attempt.

• The end of a Chargebee dunning period fires two independent decisions: one on the subscription (cancel or keep active), one on the invoice (mark not paid, void, write off, or reverse with a credit note).

• Recurly recommends "no longer than 27 days" for monthly subscriptions, which stops a second unpaid invoice stacking on the first.

• Recurly's dashboard measures invoice recovery rate, revenue recovered, and subscriptions saved from overdue invoices.

What does a dunning sequence send at each step?

A dunning sequence sends a warning before the charge, a notice after each failure, and a closing notice naming the consequence and its date. Those steps run on merchant policy, not on the card network rules governing the payment retry schedule underneath:

Stripe covers the pre-dunning end and leaves the ladder to you. Its customer email settings fire an expiring card notice "1 month before your customer's card on file expires", a trial reminder at seven days, and a failed payment email "after each failed payment". Those are separate switches, and the recovery link inside them dies once the subscription turns cancelled, incomplete_expired, or unpaid.

Ownership is why sequences drift. Engineering holds the retry cadence, support the copy, finance the suspension date, and nobody owns the whole ladder. Give one person the calendar.

When should a dunning sequence restrict access instead of sending another notice?

Restrict access once notices have stopped producing payments and the account still consumes the product. Zuora splits dunning into an invisible retry phase, a nudge phase, and an escalation phase on entitlements, triggering "service degradation" or a "service suspension" until payment arrives.

Escalation gives you more moves than an on-off switch:

• Degrade the plan. Drop to free-tier limits, keep the data reachable. Feature gating holds those limits already, so escalation is a tier change.

• Block new consumption. Stop metered spend, leave history and exports readable.

• Suspend. Access stops, the balance stays collectible, tenure survives.

• Cancel. Recurly's default marks the invoice failed at the cycle's end, which is where the account becomes involuntary churn.

Flexprice is enterprise-grade, open source usage based billing infrastructure for AI and SaaS companies. It can be deployed in your own VPC, on-prem, or on Flexprice's managed cloud. Recovery runs natively across Stripe, Razorpay, Moyasar and Nomod, so a sequence tuned on one gateway survives adding a second payment rail for India, MENA, or the GCC. Overages resolve on the same switch, charging the payment method or blocking further usage, which puts the consumption decision where the collection decision sits. The Billing and Invoicing page shows how invoices, credits, and manual payments fit together, and if your policy spans more than one payment rail, book a demo.

Which numbers show whether a dunning sequence is recovering money?

Invoice recovery rate is the headline number and it needs two companions to be readable. Recurly's dunning dashboard reports "the recovery rate of invoices that went past due during the selected date range", the revenue recovered, and "the number of subscriptions saved from overdue invoices". Each hides something alone:

• Invoice recovery rate counts invoices, so a hundred recovered $9 invoices outrank one lost $40,000 contract.

• Revenue recovered fixes the weighting and ignores how many logos you kept, while subscriptions saved can climb as dollars fall.

• Days to recovery is the one no dashboard leads with, and it says whether step three works.

Attempt count is a poor proxy for effort. Chargebee's dunning settings let Smart Retry "retry up to 12 times to collect the payment" while custom retries cap at 5, and the emails run independently of both, so comparing two configurations often compares attempt counts, not ladders. Money lost at the end lands in revenue leakage until the subscription closes.

Related terms

What a failed payment touches on its way through a dunning sequence:

• Payment retry covers the card-side mechanics: decline codes, network caps, account updater.

• Involuntary churn is what an unrecovered sequence becomes once the subscription closes.

• Subscription lifecycle holds the states dunning moves an account through.

• Feature gating is what escalation borrows when you degrade a plan instead of suspending it.

• Revenue leakage is the broader bucket unrecovered invoices fall into.

• Draft invoice is what an unpaid subscription keeps generating and never finalizing.

FAQ

Is dunning management the same as payment retry?

No. Payment retry is one component inside dunning management. Retry decides whether to reattempt a declined charge, governed largely by card network rules and decline codes. Dunning wraps those attempts in notices, an escalation path, and a final action on the subscription.

How long should a dunning period run?

Recurly recommends "no longer than 27 days" for monthly subscriptions, which keeps the sequence inside one billing period. Chargebee defines the dunning period as the days "during which an invoice will stay in dunning before the final action is carried out." Past one cycle, a second invoice generates while the first is unpaid.

What is pre-dunning?

Pre-dunning is outreach sent before a payment fails, aimed at cards about to stop working. Stripe's expiring card notice is the standard example, sending a month ahead of the expiry date. It beats recovery because nothing has failed yet.

What's the difference between dunning and collections?

Dunning runs automatically inside the billing system and targets self-serve recovery, usually a card the customer can replace. Collections is a human accounts receivable process for invoiced contracts: direct contact, payment plans, eventually an agency. Enterprise accounts on net terms often skip dunning.