An iGaming CRM migration moves the programme as well as the platform: audience definitions, triggers, permissions, journey state, reward entitlements, and measurement history. Document those first, compare the new system in shadow mode, then transfer execution in controlled tranches.

The central cutover rule is one authorised live execution path for each campaign or entitlement. Both platforms may receive events for comparison, but they must not independently send the same message or grant the same reward. Release criteria and a workable rollback procedure belong in the plan before the first live cohort moves.

What problem should the migration solve?

State the reason in terms that can be tested. A platform change may address a capability gap, a commercial arrangement, a market requirement, or an operating model the team can no longer maintain. Those reasons imply different scope and acceptance criteria.

Reason for movingEvidence to establishScope implication
Missing journey capabilityA required use case the current stack cannot meetDemonstrate that use case before selection
Integration or data limitsMeasured gaps in required events, freshness, or reconciliationPrioritise data contracts and end-to-end tests
Commercial or ownership changeDocumented contract, data access, or exit constraintsPlan overlap, export, and dependency ownership
Programme is undocumentedUnexplained audiences, restrictions, and configurationInventory and decisions before rebuilding

Do not assume replacing the tool will resolve a missing owner, unreliable event feed, or unclear lifecycle definition. Keep programme redesign bounded and explicit; otherwise nobody can distinguish migration defects from deliberate behaviour changes.

Which parts of the programme need explicit migration checks?

Trigger timing and event meaning

Measure from the originating event to the final action, including queues, eligibility checks, and downstream APIs. A platform's ingestion timestamp is only one part of that path.

Record event identity, occurrence and receipt times, timezone, schema version, ordering assumptions, and retry behaviour. Test delayed, duplicate, out-of-order, reversed, and missing events. A journey that delivers successfully can still be wrong if it acts on stale state.

Permissions, restrictions, and exceptions

Inventory consent, self-exclusion, restrictions, player-protection controls, channel preferences, contact caps, and campaign-specific exclusions. Record the reason and responsible owner for each rule.

An undocumented restriction is a review task, not automatic permission to delete it. Preserve it until the responsible owner approves its treatment. Check status changes that occur after audience selection but before execution.

Journey and reward state

Profiles alone do not preserve the programme. Capture active journey positions, timers, cooldowns, prior contacts, offer eligibility, issued entitlements, grants, expiries, and any related liabilities that the new path needs to respect.

Decide how in-flight journeys will finish, restart, or be retired. Restarting everyone at step one can duplicate messages and rewards even when the underlying player data migrated correctly.

History and measurement

Choose which history is migrated, retained in an accessible archive, or recomputed from an authoritative data source. Assess each segmentation or model dependency before deciding what can be omitted.

Preserve experiment assignments, outcome definitions, and reporting versions. An interrupted holdout or a changed contribution definition can break comparability even if campaign delivery is correct. Use the CRM KPI definitions as a reconciliation checklist.

What is a practical migration sequence?

  1. Inventory the live programme. Record journeys, audiences, templates, channels, grants, integrations, schedules, restrictions, and owners. Include external scripts and manual exports.
  2. Decide what to retain, redesign, or retire. Get the relevant owner's approval. Keep a record of intentional changes so they are not mistaken for migration failures.
  3. Define data contracts and acceptance tests. Specify event semantics, identifiers, freshness, reversals, state precedence, and financial reconciliation.
  4. Rebuild the intended logic. Translate lifecycle and eligibility rules rather than copying settings whose meaning changed between platforms. The segmentation guide provides an audience specification.
  5. Backfill and reconcile. Check counts, sample memberships, consent/restrictions, in-flight state, and financial totals under documented tolerances. Explain differences instead of accepting a blended match.
  6. Run shadow comparisons. Feed the new platform without authorising live sends or reward writes. Compare intended decisions, timing, and outputs over representative conditions.
  7. Approve a bounded pilot. Choose a representative, observable cohort and limited journey scope. Base risk on action consequences, integration complexity, and reversibility—not player value alone.
  8. Transfer execution with a recorded boundary. Fence the old path, reconcile pending actions, grant authority to the new path, and verify no overlap or gap. Expand only after pilot acceptance.
  9. Observe, reconcile, and close. Check technical execution, player experience, costs, and outcomes. Retire the old path only when rollback and retention requirements are satisfied.

Prepare a protected backup or export of the smallest complete rollback unit before changing execution. Include configuration, state, relevant data, and the authority mapping; a template export alone is not a rollback plan.

See the CRM Growth Audit

How should shadow running and cutover work?

Parallel observation is useful; duplicate execution is dangerous. The new platform can evaluate the same events and produce a dry-run decision log while the incumbent remains the sole live executor for that scope.

Compare expected audience membership, exclusions, journey steps, contact limits, intended rewards, and event-to-decision timing. Use identical snapshots where needed so a moving data feed does not create unexplained differences.

During staged cutover, platforms may each own different explicitly partitioned journeys or cohorts. They must not both own the same live action. Shared contact limits and reward eligibility still need coordination across those partitions.

Release gates to set before the pilot

GateEvidence required
Audience and status parityExplained membership differences; verified critical exclusions
Trigger timingEnd-to-end latency within use-case-specific limits, including backlog recovery
Contact controlNo duplicate or conflicting scheduled/triggered sends in the tested scope
Reward executionOne grant per intended entitlement under replay, timeout, and retry tests
Financial and state reconciliationApproved totals, liabilities, active positions, and exceptions
Operational readinessMonitoring, named owners, pause controls, and rehearsed rollback

Set numeric tolerances where appropriate from the use case and consequences. Do not accept unexplained critical exclusion failures merely because overall audience overlap is high.

Make reward retries safe

Use a stable entitlement or grant identifier enforced by the authoritative bonus system. A duplicate event, API timeout, or repeated journey step must not create a second reward. Record requested, accepted, rejected, and uncertain outcomes so reconciliation can resolve partial failures.

A successful test in the journey builder is insufficient if the downstream grant endpoint can still write twice. Test the entire path.

Rollback must account for actions already taken

Define rollback triggers, the decision owner, the last safe state, and how execution authority returns to the incumbent. Pause the new path before restoring the old one. Reconcile messages already sent and rewards already granted so restarting the incumbent does not repeat them.

Restore configuration only after accounting for post-cutover state changes. You cannot unsend a message or safely assume that restoring yesterday's database reverses a reward already used. Some incidents require a forward correction and operational response rather than replaying old state.

What should you verify in the destination platform?

Turn the programme inventory into a proof-of-capability brief. Ask the vendor or implementation team to demonstrate your difficult cases using representative data:

  • delayed and replayed events, including changed account status;
  • multi-brand and market boundaries;
  • journey exits, cooldowns, and in-flight state transfer;
  • central contact controls and reward deduplication;
  • persistent holdouts and exportable assignment/outcome records;
  • operator access, audit history, incident controls, and data export on exit.

Compare the implementation work and ongoing operating cost, including channel dependencies and overlap. A platform category alone does not establish integration effort or fit.

The CRM platform evaluation framework addresses the broader decision. The features and integration checklist turns that decision into practical evidence requests. For an independent view of the plan, see CRM technology advisory.

Frequently asked questions about iGaming CRM migration

How long does a migration take?

Scope it from the inventory, data contracts, integrations, in-flight state, review dependencies, and agreed test coverage. A generic duration cannot account for those differences. Define milestones by accepted evidence, not only calendar dates.

Do both platforms need to run live?

They can receive data and calculate decisions concurrently. For each live scope, retain one authorised execution path. Shadow comparison should not produce a second message or bonus grant. If scopes are partitioned during rollout, document the boundary and shared controls.

Should you migrate historical data?

Keep the history required by journeys, permissions, models, reporting, and applicable retention obligations. Decide whether it transfers, is recomputed, or remains accessible in an archive before cutover. More history is not automatically better; use and retention should be defined.

Is matching the old platform enough?

It establishes parity only for the observed cases. Test failure handling and changed eligibility too. If logic is intentionally redesigned, document the difference and evaluate its effect separately rather than treating every mismatch as an error.

How do you measure the business effect of migration?

Compare execution and experience first, then compatible cohorts and financial outcomes by product and market. A simple before/after change does not isolate migration from seasonality or other changes. Use an appropriate experiment where feasible; the campaign ROI guide explains the distinction.

When can the old platform be retired?

After accepted release checks, reconciliation of in-flight actions and liabilities, expiry of the agreed rollback window, and confirmation of required exports and retention. Remove obsolete credentials, jobs, and connections under a documented closure checklist.