How to switch acquirers without downtime

Acquirer migration is one of the most misunderstood operational projects in payments. Operators assume "we're switching from X to Y" means a clean cutover. In reality, a zero-downtime migration involves 60-90 days of parallel operation, careful subscription token handling, chargeback tail management, and explicit cutover of multiple downstream systems. Done right, customers don't notice. Done wrong, you lose subscribers and eat chargebacks on accounts that are no longer accepting them.

Pre-migration assessment

Understand what's moving

• Transaction processing (obvious)
• Stored payment tokens (subscriptions)
• Chargeback / dispute flow
• Webhook delivery endpoints
• Reconciliation / accounting feeds
• Fraud tool integrations
• Reserve balance (stays with old acquirer during hold)

Understand what stays

• Customer data (CRM, email, etc.)
• Product catalog
• Checkout UX (most of it)
• Customer login + portal

The parallel-processing window

Don't cut over on day one. Run both acquirers in parallel.

Week 1 — new acquirer live for new traffic

• New transactions route to new acquirer
• Existing subscriptions continue on old acquirer
• Monitor both for anomalies

Week 2-4 — subscription migration begins

• Network tokens migrate from old to new acquirer where possible
• For tokens that can't migrate, trigger re-auth flow
• Customer sees "update payment for your subscription" email
• Customer clicks, enters card on new acquirer's hosted form, subscription transitions

Week 5-8 — remaining subscription cleanup

• Remind customers who haven't migrated
• Process last charges on old acquirer for unmigrated subs
• Offer small incentive (extended delivery, small discount) to customers who migrate proactively

Week 9-12 — old acquirer wind-down

• No new charges on old acquirer
• Chargeback tail still processing
• Reconciliation closes per old schedule
• Reserve release tracking

Subscription token migration — the hard part

Network tokenization migration

Visa Token Service and Mastercard MDES allow inter-acquirer token portability in some cases. Requires:

• Both acquirers enrolled in VTS/MDES
• Coordination between acquirers for token handoff
• Cardholder consent (sometimes)
• Old acquirer's cooperation (they may not want you to leave clean)

When network token migration works

• Both acquirers modern and cooperative
• Platform supports VTS/MDES fully
• Customer authorized in terms

When it doesn't

• Old acquirer doesn't cooperate
• Tokens are acquirer-proprietary (not network tokens)
• Card types not covered

Re-auth fallback

Customer receives email: "We're updating our payment system. Please re-enter your card to keep your subscription active."

• Link to hosted card form on new acquirer
• Customer enters card, subscription transitions
• Expected success rate: 60-80% over 30 days
• Expected churn: 20-40% (customers who don't act)

Chargeback tail

Charges on old acquirer can chargeback for 120 days. Migration doesn't end chargeback responsibility.

• Keep dispute team on old acquirer's dashboard for 180 days post-last-charge
• Reserve holds on old acquirer cover outstanding chargebacks
• Reserve release after chargeback window closes
• Expect 10-30% of final month's charges to chargeback (elevated vs steady state)

Webhook / integration cutover

Dual webhooks during parallel window

• Old acquirer webhooks continue
• New acquirer webhooks live in parallel
• Downstream systems (CRM, accounting, fulfillment) idempotent to handle duplicate events

Final cutover

• When no more new charges on old acquirer, disable old webhooks
• Monitor new webhooks for drops
• Reconcile any gaps

Reconciliation during parallel period

Two processor feeds to reconcile. More complex than normal:

• Tag each charge with originating acquirer
• Reconcile gross per acquirer
• Reconcile fees per acquirer (different rate structures)
• Combine for management reporting

Reserve transition

• Old acquirer reserve releases per schedule (typically 180 days post-last-charge)
• New acquirer reserve starts accruing on new volume
• Total working capital tied up spikes temporarily (both reserves active)
• Plan treasury accordingly

Customer communication

Strong customer comm is the difference between clean migration and subscriber loss.

Proactive heads-up (2 weeks before migration)

• "We're upgrading our payment system"
• "You may see a small change in descriptor"
• "Your subscription continues uninterrupted"

If re-auth needed

• Clear email: "Please re-enter your card to keep [PRODUCT] active"
• Make it easy (one-click from email)
• Small incentive for quick action

Post-migration confirmation

• "Your subscription is fully active"
• "New descriptor on your statement will be X"
• "Support contact unchanged"

Multi-brand migration

Portfolio migration is harder than single-brand. Recommended approach:

• Migrate brand-by-brand over 3-6 months
• Start with smallest brand to validate process
• Progress to larger brands after validation
• Keep reconciliation disciplined across brands × acquirers during parallel

Fraud tool migration

Sift, Signifyd, Kount integrations often need reconfiguration:

• Old acquirer identifiers no longer valid
• New acquirer integration may differ
• Historical fraud data needs to port

Compliance during migration

• PCI attestation on new acquirer before first live transaction
• Updated BAA if HIPAA-adjacent
• Updated data-processing agreements per state/country regulation
• Updated vendor list in SOC 2 if applicable

Rollback scenarios

What if migration goes wrong?

• Revert new-transaction routing to old acquirer
• Tokens already migrated: re-auth flow the other direction (more friction)
• Chargebacks on both accounts until cleaned up
• Rollback is painful but possible during parallel window

Post-cutover rollback is much harder. Decide on full cutover only with confidence.

Timeline summary

• Week -4 to 0: planning, underwriting on new acquirer, integration dev
• Week 1: new acquirer live for new traffic
• Weeks 2-4: subscription migration begins
• Weeks 5-8: subscription migration cleanup
• Weeks 9-12: old acquirer wind-down
• Weeks 13-24: chargeback tail, reserve release

What not to do

• Don't flip switch without parallel window — subscription loss will be significant.
• Don't skip customer communication — silence triggers disputes.
• Don't let dispute team quit on old acquirer at cutover — chargeback tail is long.
• Don't migrate during peak season — do it during slow months.

What to do next

Plan the migration 90 days before go-live. Budget 20% subscription churn as realistic worst-case on re-auth migration. Monitor every metric for 30 days post-cutover.

Multi-brand portfolio migrations: our application covers structured migration planning. See also 30-day cutover plan.