Processor Switch Checklist (Playbook)
Switching processors is high-stakes. Mess up the migration and you lose recurring revenue, break checkout, or create reconciliation nightmares. This playbook keeps the transition clean.
Timeline: 4-8 weeks minimum. Complex migrations take 12+ weeks.
Workflow Overview
| Phase | Key Tasks |
|---|---|
| Discovery | Inventory current processor, map features to new, define token strategy |
| Integration | Build parallel integration, test matrix (auth, capture, refund, void), all flows |
| Token Migration | Export strategy, import tokens, validate mapping, cutover planning |
| Execution | Ramp 10% → 25% → 50% → 100%, monitor auth rate, migrate recurring billing |
| Cleanup | Sunset old processor, update docs, post-migration review |
Prerequisites: New processor contract signed, test credentials, token export policy known, engineering resources allocated.
Trigger Criteria
Use this playbook when:
- You're migrating to a new payment processor
- You're adding a backup processor
- You're consolidating from multiple processors to one
- Current processor relationship is ending
Prerequisites
Before starting:
- New processor contract signed
- New processor account approved and live
- Test credentials for new processor
- Current processor token/card data export policy known
- Engineering resources allocated
Phase 1: Discovery (Week 1-2)
Step 1: Inventory Current State
Document everything about current processor:
| Item | Document |
|---|---|
| Transaction types | One-time, recurring, auth+capture, etc. |
| Payment methods | Cards, ACH, wallets, local methods |
| Token format | Vault structure, token format |
| Recurring billing | Subscription count, billing frequency |
| Stored credentials | How many cards on file |
| MIDs | All merchant IDs, per-MID config |
| Reporting | Settlement, reconciliation, disputes |
| Integrations | APIs, webhooks, SDKs in use |
Step 2: Map to New Processor
| Feature | Current | New Processor | Gap? |
|---|---|---|---|
| Card vault | Yes/No | Yes/No | |
| Network tokens | Yes/No | Yes/No | |
| 3DS | Version | Version | |
| ACH | Yes/No | Yes/No | |
| Apple/Google Pay | Yes/No | Yes/No | |
| Reporting format | Format | Format |
Checkpoint: Feature parity confirmed or gaps documented with workarounds.
Step 3: Token Migration Strategy
| Scenario | Approach |
|---|---|
| Processor supports token migration | Export tokens, import to new processor |
| No direct migration | Use network tokens (Visa/MC token services) |
| Neither works | Re-collect cards at next transaction |
| Account updater available | May help with expired cards during transition |
Critical question: Can you export raw PANs or tokens that the new processor accepts?
Phase 2: Integration (Week 2-4)
Step 1: Build New Integration
Parallel integration approach:
Step 2: Integration Checklist
- Auth request/response handling
- Capture flow
- Refund flow
- Void flow
- Partial capture (if used)
- 3DS integration
- AVS/CVV handling
- Error handling and retries
- Webhook receivers
- Tokenization flow
- Recurring billing integration
Step 3: Test Matrix
| Test Case | Status |
|---|---|
| Successful card auth | |
| Declined card (various codes) | |
| 3DS challenge flow | |
| 3DS frictionless flow | |
| Auth + capture | |
| Auth + void | |
| Auth + partial capture | |
| Refund (full) | |
| Refund (partial) | |
| Recurring billing (create) | |
| Recurring billing (charge) | |
| Recurring billing (update) | |
| Apple Pay | |
| Google Pay |
Checkpoint: All test cases passing in sandbox.
Phase 3: Token Migration (Week 3-5)
Option A: Processor-to-Processor Migration
If new processor accepts imported tokens:
- Request token export from current processor
- Transform token format if needed
- Import tokens to new processor vault
- Map old token IDs to new token IDs
- Update your database with new token references
- Validate sample of migrated tokens
Option B: Network Token Migration
Using Visa/Mastercard network tokenization:
- Ensure both processors support network tokens
- Export network tokens (DPAN + cryptograms) from old processor
- Register network tokens with new processor
- Network tokens work across processors
Option C: Gradual Re-collection
If migration not possible:
- Keep old processor active for existing tokens
- New customers go to new processor
- At next transaction, existing customers re-enter card
- Migrate stored credential to new processor
- Eventually sunset old processor
Checkpoint: Migration approach chosen and tested with sample data.
Phase 4: Cutover Planning (Week 4-6)
Step 1: Define Cutover Strategy
| Strategy | Pros | Cons |
|---|---|---|
| Big bang | Clean cutover | High risk |
| Percentage ramp | Lower risk | Longer timeline |
| New customers only | Lowest risk | Two systems forever |
| By segment | Targeted migration | Complexity |
Recommendation: Percentage ramp (10% → 25% → 50% → 100%)
Step 2: Create Rollback Plan
Define criteria and process:
| Trigger | Action |
|---|---|
| Auth rate drops 5%+ vs baseline | Pause ramp, investigate |
| Error rate over 1% | Route back to old processor |
| New processor outage | Automatic failover to old |
| Dispute rate spikes | Investigate before proceeding |
Step 3: Communication Plan
| Stakeholder | What to Communicate | When |
|---|---|---|
| Customer support | New error codes, escalation path | 1 week before |
| Finance | New settlement schedule, reconciliation changes | 2 weeks before |
| Engineering | On-call schedule, rollback procedures | 1 week before |
| Customers | Only if checkout UX changes significantly | Optional |
Phase 5: Migration Execution (Week 5-8)
Step 1: Initial Ramp (10%)
- Route 10% of new transactions to new processor
- Monitor for 3-5 days minimum
- Check:
- Auth rate vs baseline
- Error rate
- Customer complaints
- Settlement accuracy
Step 2: Expand Ramp (25% → 50% → 75%)
At each stage:
- Increase traffic percentage
- Monitor for 2-3 days
- Verify metrics stable
- Proceed or troubleshoot
Step 3: Full Cutover (100%)
- Route 100% to new processor
- Keep old processor available for fallback
- Monitor closely for 1 week
- Begin old processor sunset planning
Step 4: Recurring Billing Migration
Handle separately from one-time payments:
- Identify all active subscriptions
- Migrate tokens (per Phase 3 approach)
- Update subscription records with new tokens
- Test billing on subset of subscriptions
- Run first full billing cycle with monitoring
- Validate all charges succeeded
Checkpoint: All traffic on new processor, metrics stable.
Phase 6: Cleanup (Week 6-10)
Step 1: Sunset Old Processor
- Confirm no active traffic to old processor
- Resolve any pending transactions/settlements
- Download historical data for records
- Cancel old processor contract (if applicable)
- Remove old processor code paths (after stabilization)
Step 2: Update Documentation
- Internal runbooks updated
- Dispute handling procedures updated
- Reconciliation processes updated
- Alert thresholds calibrated for new processor
Step 3: Post-Migration Review
| Question | Answer |
|---|---|
| Auth rate vs pre-migration? | |
| Any lost recurring customers? | |
| Settlement timing acceptable? | |
| Support ticket volume? | |
| Lessons for next migration? |
Success Criteria
You're done when:
- 100% of traffic on new processor
- Auth rate within 0.5% of pre-migration baseline
- All recurring billing functioning
- Reconciliation process working
- Old processor dependencies removed
Scale Callout
| Volume | Considerations |
|---|---|
| Under $100k/mo | Can often do big-bang cutover; simpler token migration |
| $100k-$1M/mo | Percentage ramp recommended; test recurring billing carefully |
| Over $1M/mo | Dedicated migration team; extended parallel running; fallback essential |
Where This Breaks
- Token migration fails. Test migration on small sample first. Have re-collection fallback.
- Auth rate tanks. New processor may have different issuer relationships. Compare BIN-level before committing.
- Recurring billing broken. Test billing flow end-to-end before migrating subscriptions.
- Settlement reconciliation off. Different settlement timing/format. Update finance processes.
- Old processor shuts down early. Get written timeline. Keep old processor available longer than planned.
- No rollback tested. Test failover before you need it. Don't assume it works.
Critical Metrics to Track
| Metric | Pre-Migration | During Ramp | Post-Migration |
|---|---|---|---|
| Auth rate | Baseline | Daily | Weekly |
| Error rate | Baseline | Hourly | Daily |
| Decline codes | Distribution | Daily | Weekly |
| Settlement lag | Baseline | Daily | Weekly |
| Dispute rate | Baseline | Weekly | Monthly |
Next Steps
After migration is complete:
- Optimize on new processor: Run auth optimization experiments now that you have fresh baseline
- Clean up old code: Remove old processor integration after 30-day stability period
- Update documentation: Runbooks, escalation contacts, API references
- Renegotiate terms: At 6-month mark, review volume and negotiate better rates
- Plan redundancy: Consider keeping a backup processor relationship for failover
Related
- Auth Optimization - Improving approval rates
- Processor Reporting Checklist - Data requirements
- Settlement & Reconciliation - Post-capture operations
- Benchmarks - Auth rate targets
- Processor Management - Working with processors
- Subscriptions & Recurring - Migrating recurring billing
- 3D Secure - Authentication integration
- Decline Codes - Understanding declines
- Holds and Reserves - Processor reserves
- Payment Change Checklist - Testing changes
- Reading Statements - Understanding processor costs
- Buying Payments - Processor selection criteria