Dunning is often described as sending reminders, but a reliable subscription workflow is a sequence of account states. It must distinguish a failed payment from a disputed charge, expired card, procurement delay, suspected fraud, technical billing error, and deliberate cancellation. A Philippines-based billing coordinator can run the documented sequence and prepare exceptions while the client retains pricing, credit, refund, suspension, and write-off authority.
Start from account states, not email cadence
Map the states that exist in the billing platform: current, payment retry pending, customer action needed, internal correction needed, dispute open, promise to pay, grace period, restricted, cancelled, and closed. Define the event that enters and exits each state. A calendar saying “send three emails” is insufficient because a successful retry, open support incident, or approved procurement extension should change the path.
For every state, name the permitted message, system action, evidence, owner, timer, and escalation. Make timers use one declared timezone and show whether weekends count. If an automation changes status, preserve the triggering event and template version. The offshore coordinator should be able to explain why an account received a message without reverse-engineering an opaque campaign.
Clean the failure signal before contacting customers
Payment gateways return reason codes with different meanings and disclosure limits. Convert them into internal action groups approved by the payment owner: customer can update details, retry may be appropriate, merchant configuration needs review, or specialist investigation is required. Do not expose raw risk signals or imply that a bank rejected a customer for a specific reason unless the approved message supports it.
Suppress outreach when the invoice is already paid, a credit covers the balance, the subscription is cancelled, a dispute is being investigated, or a documented technical incident caused the failure. Reconcile gateway, billing, and customer-success data before the first manual touch. A wrong reminder damages trust more quickly than a delayed but accurate one.
Give the coordinator a bounded action set
Permitted tasks may include verifying contact details from an approved source, checking invoice delivery, selecting an approved reminder, recording the customer’s response, resending a secure payment-update link, and routing a complete exception packet. Prohibit staff from collecting card numbers in email or chat, changing prices, extending service, promising fee waivers, issuing credits, or bypassing authentication.
Permissions should reflect those limits. A coordinator who only prepares cases does not need refund or plan-change rights. Use named accounts, multifactor authentication, and audit logs. Temporary access should expire automatically. When a customer requests an account change, follow the organization’s authentication procedure rather than treating possession of an email thread as proof of authority.
Write messages for useful action
Each reminder should identify the account safely, explain the current status without blame, provide the approved secure action, state a truthful deadline, and offer a route for disputes or accessibility needs. Avoid false urgency. Do not say an account “will be deleted tonight” unless that is the configured and authorized outcome.
Localize time and currency carefully. The operations team may work in the Philippines while customers, billing entities, and payment processors operate elsewhere. Templates need an explicit date, timezone, currency, and tax context. Test links in a non-production environment and confirm that tracking parameters do not leak account information.
Route exceptions by cause
Create separate lanes for payment-method update, invoice or tax-document request, billing defect, contractual dispute, suspected fraud, customer hardship, procurement delay, and cancellation request. Each lane needs a decision owner. The coordinator can assemble facts, but a customer asking for a different price or extra grace is requesting a commercial decision.
Consider a business customer whose card failed while its annual purchase order is awaiting approval. The coordinator verifies the invoice destination, records the procurement date supplied by the authorized contact, and routes an extension request with account history. The account owner decides whether service continues. The record makes clear that the coordinator did not grant credit.
Coordinate retry automation and human contact
Automated retries and manual outreach can collide. Publish retry times to the team, prevent duplicate messages, and refresh account status immediately before a manual action. If the customer updates a method, the workflow should wait for a confirmed processor result rather than assuming the form submission succeeded.
Use idempotent actions where the platform supports them. Repeated clicks should not create several invoices or retries. Give staff a recovery instruction for timeouts: check transaction status before attempting again. This small rule prevents many duplicate charges and confusing customer conversations.
Treat disputes and cancellations as different work
A disputed transaction belongs in a documented dispute process with preserved communications and processor deadlines. A cancellation is a request to stop future service under the applicable terms. Neither should be buried in a generic failed-payment note. Train the team to recognize phrases that trigger these routes and to acknowledge them promptly without arguing the merits.
Retention offers, refund decisions, and interpretations of contract terms remain with authorized staff. The offshore role may schedule a callback, assemble the subscription history, and confirm that the request reached the right owner. The closure reason must reflect the actual customer outcome, not whatever clears the queue fastest.
Use measures that expose harm
Useful measures include incorrect-contact rate, suppression failures, successful secure updates, accounts requiring internal correction, exception age, duplicate outreach, and customer complaints linked to dunning. Recovery rate matters, but it should not outweigh accuracy or fair treatment. Segment results by failure group and account type before drawing conclusions.
Review a sample of messages against the account state that existed at send time. Check that the template, deadline, link, and escalation were correct. Analyze false positives separately: contacting a paid or cancelled customer is a control failure even if no money was lost.
Launch with a shadow period
First let the coordinator classify historical failures and prepare messages without sending them. Compare classifications with the billing owner, revise ambiguous rules, then enable a narrow customer segment with daily review. Test expired links, repeated webhook delivery, processor outage, disputed invoices, and changes arriving just before suspension.
The PCI Security Standards Council publishes standards and guidance for payment-account data. The FTC’s business guidance offers consumer-protection resources, and the NIST Cybersecurity Framework provides a governance structure for access and incident handling. Apply the actual processor contract, customer terms, laws, and qualified advice to the operation.
Good dunning makes the next legitimate action obvious and prevents inappropriate ones. A Philippines team can operate that system consistently when account states, secure channels, customer language, and retained decisions are designed before volume arrives.