For an established Australian training consultancy
Move the systems. Keep the promises you have already made.
You have a cohort starting next week, coaching sessions already booked and clients who expect their learning history to be there when they sign in. Consolidating software could make your Melbourne training consultancy easier to run. That does not make an unexplained login change on a Monday morning acceptable.
The useful question is: what must remain true for each client while you move? A paid learner should retain the access you promised. A booked appointment should retain its correct time. Someone who declined marketing should not suddenly receive a launch sequence. These are the acceptance checks around which to build your software migration plan.
Use the planner above to sketch the scale of the work. Its three inputs produce a phased checklist and a rough two-to-ten-week starting estimate. It does not inspect your systems, transfer records or verify that a payment arrangement can move. Treat the output as an agenda for investigation, then use the decisions below to turn it into your own runbook.
Before choosing a launch date, answer these five questions
For your next live cohort, choose one record that exposes a difficult case rather than your easiest customer.
- Can you identify the authoritative enrolment, access-expiry date and any completion evidence?
- Can you separate the learner from the employer who purchased their place?
- Can you preserve booking time, location and remaining session entitlement?
- Can you identify the latest marketing opt-out and prevent import-triggered messages?
- Can an authorised operator recover service if a rehearsal fails?
An unknown answer becomes an assigned investigation with a named owner. Do not turn it into an assumption on the launch checklist.
This is an operational diagnostic, not an automated audit of your accounts or a legal compliance assessment.
- 1 · Map records and dependencies
- 2 · Rehearse a representative sample
- 3 · Reconcile and test recovery
- 4 · Approve or postpone the move
An unresolved acceptance check returns to investigation. The calculator is a planning heuristic; it does not perform these checks.
1. Inventory client promises before counting applications
Start with the delivery journey: purchase, enrolment, preparation, attendance, follow-up and renewal. Against each step, record the source system, destination, record identifier, responsible person and what would fail if the move went wrong. This exposes dependencies a subscription list misses.
Keep historical evidence and operational access separate. You may need an archived completion record without recreating an expired enrolment. Decide what must remain accessible, who can access it and when you will review retention. Your business obligations need a separate check; this guide does not assign a universal retention period.
- Why it matters
- A dependency map follows the relationship between records, so a successful file import cannot conceal a broken learner entitlement.
- Work through the example
- An employer buys ten places, but only eight learners have registered. Preserve the purchase-to-cohort relationship and two unused places; ten new learner accounts would misrepresent the situation.
- When the approach changes
- If you are moving only a public resource library, this depth of enrolment mapping may be unnecessary. Still check incoming links and ownership of the files.
- Your decision
- Choose one authoritative source for each entitlement before mapping it into the destination.
2. Rehearse the awkward records, not just the clean ones
A useful sample includes an active learner, a completed learner, an expired enrolment, a duplicate email, an employer-funded place and a client with a future coaching appointment. Add your own exceptions. Run the rehearsal in a controlled environment with customer messages disabled and authorised access to the data.
One documented course-platform import service illustrates why the actual transfer route matters: its described service cannot carry arbitrary partial progress, and its notification and event behaviour has specific limits. Those details are evidence about that service, not promises about Gbeya. Ask for field-level evidence for your exact source and destination.
- Why it matters
- Sample reconciliation tests meaning as well as row counts: who can access what, until when, and with which history.
- Work through the example
- Two exports may both contain 240 learners while one import gives everyone indefinite access. Compare expiry and entitlement fields before calling those totals a match.
- When the approach changes
- A clean sample does not establish that a large export is complete. Keep whole-file counts, rejected-row reasons and exception reconciliation too.
- Your decision
- Do not schedule cutover until every critical sample case passes or has an explicit, workable exception agreed by its owner.
3. Give every message and booking one active owner
Write down which system sends each reminder, welcome message and follow-up. During overlap, avoid letting both systems interpret an imported enrolment as a new sale. Keep a record of the last processed event and the point at which responsibility moves.
Preserve email status separately from customer status. A paying client can still be opted out of marketing. For your Australian delivery schedule, check actual appointment dates in the client’s location: a Melbourne-to-Perth session needs a deliberate time-zone check, especially around daylight-saving changes.
- Why it matters
- Explicit ownership gives you a way to detect and avoid duplicate actions; testing status mappings helps you catch an import that changes the meaning of a contact record.
- Work through the example
- Your rehearsal learner receives one planned service-access message and no marketing welcome sequence. The booking shown to a Perth client resolves to the same instant in the coach’s calendar.
- When the approach changes
- Some operational messages legitimately come from different systems. Keep that arrangement when each message has a distinct purpose and a tested trigger.
- Your decision
- An unexplained duplicate message or changed appointment instant is a failed acceptance check, even when the import itself reports success.
4. Define the stop decision before the pressure arrives
Put the go/no-go decision in the runbook with an owner, a deadline and observable conditions. Examples include inaccessible paid learning, mismatched active entitlements or an unresolved booking conflict. Agree which defects can wait and which stop the move.
Rollback becomes harder once new records accumulate in the destination. Before switching traffic, define how you will preserve those new bookings or enrolments if you return. Keeping an old login does not solve conflicting records. Rehearse restoration and the operational handover; do not discover the procedure during a live incident.
- Why it matters
- A written decision point separates evidence from launch-day optimism and forces you to account for changes after the switch.
- Work through the example
- If the controlled access test fails, the owner keeps the existing learner portal active and postpones the new entry link. If new enrolments already exist, the recovery procedure must reconcile them before any return.
- When the approach changes
- For a small isolated workflow, fixing forward may be safer than reversing everything. Choose based on recoverability and client impact, not a blanket rollback rule.
- Your decision
- If neither recovery nor a safe forward fix has been demonstrated for a critical failure, the move is not ready.
5. Retire a system only when its remaining job is understood
Consolidation is a means to reduce operational friction. It is not a requirement to move every historical record or recurring billing arrangement on the same day. Keep an exception register: what remains, why, who owns it and what evidence would permit retirement.
Before cancelling, check export completeness, required evidence access, unresolved support cases and any still-active customer journey. Use your own renewal dates and operating costs. There is no universally correct overlap period.
- Why it matters
- A time-bounded exception protects continuity without allowing the old stack to remain indefinitely through habit.
- Work through the example
- A current cohort finishes in the existing environment while the next cohort starts in the new one. Staff have a clear routing list and review date, rather than guessing where a learner belongs.
- When the approach changes
- Overlap itself can create confusion and cost. If the workflow is simple and fully rehearsed, a single controlled switch may be cleaner.
- Your decision
- Retain only what has a defined continuing purpose; retire only after that purpose has ended or been safely transferred.
The strongest case is for tested continuity, not migration at any cost
Supplier guidance shows that moving records is feasible but route-specific. Australian security guidance supports testing recovery, while the countercase for retaining a workload is strongest when dependencies remain unresolved. An operator’s recovery account illustrates that concern but cannot tell us how often it happens.
For your consultancy, the practical conclusion is to prove the difficult client journeys before choosing the switch date. That is a medium-confidence operational recommendation across different source contexts, not evidence that one platform or migration method always wins.
Confidence: Medium. No source establishes your destination’s compatibility, your required duration or your realised savings.
Planning calculation, substitute your own figures
Budget for continuity before claiming savings
Suppose you plan 12 hours of internal work at an assigned AUD80 per hour, AUD300 of temporary software overlap and an AUD240 contingency. The planning allowance is AUD1,500: 12 × 80 + 300 + 240. If the completed move later removes AUD250 of monthly recurring costs, simple recovery would take six months, excluding tax and any other costs or benefits. The saving is an assumption until subscriptions are actually retired. Separately, suppose 240 active enrolments export and 238 reconcile: the two exceptions block acceptance until explained. A financially attractive estimate does not overrule that operational check.
If the older system must stay longer, overlap costs rise and the saving starts later. Recalculate rather than keeping the original payback claim.
Approve the move on verified compatibility and an affordable downside; measure savings after retirement.
Assumptions worth checking
If the exported and imported totals match, the move is complete.
Matching totals can hide changed access dates, missing relationships and altered message triggers. Reconcile meaning as well as quantity.
An experienced operator can simply switch back if anything goes wrong.
New records may exist only in the destination. A return needs a tested plan for those changes, not just the old account credentials.
Your next checks, with a clear acceptance decision
Days 1–7
Map and decide
Map the live cohort journey, inventory dependencies, assign record ownership and collect export specifications.
Ready to proceed: Every critical client promise has a source, destination and named reviewer.
Days 8–14
Rehearse exceptions
Test a controlled sample, reconcile permissions and dates, and check message ownership.
Ready to proceed: Critical sample failures are resolved or the move is held with reasons.
Days 15–21
Prepare the decision
Document client communication, recovery steps, change ownership and go/no-go conditions. Rehearse them.
Ready to proceed: The decision owner can explain how service continues if the switch is postponed.
Days 22–30
Switch only if ready
If acceptance passes, perform the controlled cutover and review exceptions. Otherwise continue the existing service and revise the plan.
Ready to proceed: The observed results, unresolved issues and retirement conditions are recorded; no arbitrary deadline overrides continuity.
Questions? See the Help centre
