MES Cutover With a 48-Hour Window
A contract manufacturer is six weeks from replacing its legacy MES with a new system across two plants and 40 production stations. ERP integration is built, but master-data reconciliation found 3,200 item records with an 11% mismatch rate; critical attributes such as lot control and unit of measure are affected. Operator training is 60% complete, and the plant has no experience with dual transactions. The planned cutover is a 48-hour weekend big-bang across both plants. Current OEE is 72%, and the customer expects no shipment delays. The program manager proposes proceeding because the ERP interface tested successfully in isolation. The operations director wants a go/no-go recommendation with explicit criteria.
1. Recommend go/no-go for the 48-hour big-bang cutover. Identify the decision criteria and the minimum conditions for proceeding.
Read the worked answer
Recommend no-go for the big-bang cutover across both plants. A successful ERP interface test proves message transport, not operational readiness. An 11% master-data mismatch on 3,200 items means roughly 350 records may carry wrong lot-control or unit-of-measure attributes. Those errors surface as wrong labels, incorrect inventory, or blocked shipments. Training at 60% and no dual-transaction experience leave operators learning the system while trying to protect customer deliveries.
Minimum conditions to proceed: critical item attributes reconciled to 100%, noncritical records below an agreed threshold; training above 95% with competency checks, not attendance; a tested rollback and reconciliation procedure; a pilot on one line or one plant with at least two weeks of dual reporting; and named operations ownership for cutover decisions.
If those conditions cannot be met in six weeks, move to a phased cutover rather than declaring the program late. The decision should be based on shipment and inventory risk, not on interface test results.
2. Propose a phased cutover plan and the operational KPIs that would trigger a rollback.
Read the worked answer
Cut over by value stream or plant, starting with the lowest-risk line that still exercises lot control, labeling, and ERP posting. Run legacy and new systems in parallel for a defined period, with manual reconciliation at shift end. Freeze master-data changes before each wave, and use a controlled exception process for urgent items. Train the next wave on lessons from the first.
Track leading and lagging KPIs daily: OEE against the 72% baseline, first-pass yield, work-order release and closure rate, inventory accuracy, label errors, transaction latency, manual override rate, and shipment on-time performance. Set rollback triggers in advance. Examples: OEE below 65% for two consecutive shifts, order release failures above 5%, inventory mismatch above 2% at shift close, any mislabeled lot released to a customer, or ERP posting backlog that cannot be cleared within one shift.
Rollback must restore the legacy system as the system of record, not merely stop the new system. Rehearse that step before the first wave and assign an operator, not only IT, to declare success or rollback.