Warehouse System Go-Live With 21 Days Left: Conditions, Escalation, and the Expiring Licence
Northwind Retail is scheduled to go live with a new warehouse management system across 6 distribution centres in 21 days. At the go-live readiness review: 47 open defects, of which 6 are severity 1, including a carton-label print failure that staff work around by reprinting at a supervisor station; 18 are severity 2. Data migration reconciliation shows 99.2% of SKU records matched, but the unmatched 0.8% includes 1,900 SKUs carrying 14% of inventory value. User acceptance testing was executed by the IT testing team; only 2 of the 6 sites participated. The parallel run lasted 3 days at a single site. The vendor contract caps liability at 12 months of fees. Peak season begins 45 days after go-live, and the legacy system's licence expires in 30 days. The project sponsor describes the remaining defects as cosmetic.
1. State your go-live recommendation and the conditions that would have to be met for it to hold. Justify each condition by reference to a specific fact.
Read the worked answer
Recommendation. Conditional go-live is not supportable on current evidence; the correct position is a deferral of the decision for a defined period, with a narrow set of gating conditions. If the sponsor insists on the date, the auditor should record a formal, documented objection rather than soften the finding.
Condition 1: Resolve or formally accept all severity 1 defects, with the label failure resolved rather than worked around. The workaround is a manual reprint at a supervisor station. That is a capacity and accuracy control operated by people under peak load, and it converts a system defect into a throughput bottleneck at exactly the moment volumes triple. Acceptance must be signed by the operations owner who will live with it, not by the project.
Condition 2: Complete migration reconciliation down to value, not record count. A 99.2% match sounds strong, but the 0.8% carries 14% of inventory value. Mis-stated inventory in a WMS produces incorrect replenishment and picking, and the financial statements inherit the error. Every one of the 1,900 SKUs must be individually reconciled or explicitly written off before cutover.
Condition 3: Business-led UAT at all 6 sites. Testing executed by IT answers whether the system behaves as specified; it cannot answer whether the warehouse can operate it. Four sites have never touched the system, and peak is 45 days away.
Condition 4: Extend the parallel run across at least one high-volume site through a full weekly cycle. Three days at one site does not cover month-end, payroll cutover, or carrier peak.
Condition 5: Address the licence expiry as a commercial, not a technical, event. Negotiating a short extension is cheaper than an uncontrolled cutover, and it removes the false urgency driving the timetable.
2. Which issues would you escalate beyond the project, to whom, and how would you phrase them so they are acted on rather than defended against?
Read the worked answer
Escalation should be selective. Escalating all 47 defects dilutes the message; escalating four structured items gets action.
To the audit committee and the CFO. The inventory valuation exposure: 1,900 SKUs holding 14% of inventory value are unreconciled 21 days before cutover. Phrase as a financial reporting risk with a number attached, not as a testing gap. Ask for the reconciliation plan owner and completion date.
To the CEO or COO. The peak-season sequencing risk: go-live is 45 days before peak, with four of six sites untrained and untested. The decision to proceed is a business risk acceptance, and it should be made by the executive who owns peak-season service levels, in writing. Frame it as a question of who accepts the risk, not an assertion that the project is failing.
To procurement and legal. The liability cap of 12 months of fees against a failure at peak, and the legacy licence expiring in 30 days. Both are contract positions that can still be improved before signature or renewal; after go-live the leverage is gone.
To the project sponsor. The severity 1 workaround, described neutrally: the manual reprint step, its expected volume at peak, and the staff it consumes. Sponsors respond to operational arithmetic.
Framing discipline. State the fact, the consequence with a number or a date, and the decision required from the recipient. Avoid adjectives such as cosmetic or negligent; they invite argument about language instead of action. Put the auditor's own recommendation in writing and retain it, because if the project proceeds and fails, the quality of the escalation will be judged afterwards.
3. Design the post-implementation review: when should it start, what should it cover, and what would make it worthless?
Read the worked answer
Timing. Start after at least one full peak cycle and one financial close on the new system, so roughly four to six months after go-live. A review run at 30 days captures stabilisation noise and cannot see whether inventory accuracy converged. However, benefits tracking should be baselined before go-live, because the pre-implementation metrics are the only valid comparison.
Scope.
- Benefits realisation against the original business case: order accuracy, lines picked per hour, inventory accuracy, and cost per unit shipped, measured on the same definitions used pre-implementation.
- Defect burn-down: whether severity 1 and 2 counts actually fell, or were reclassified to clear the report. Compare the defect register before and after go-live.
- Data quality: whether the 1,900 unmatched SKUs were resolved, and whether new discrepancies are being created at a stable or growing rate.
- Control performance in the new environment: segregation of duties in the WMS roles, interface reconciliation between WMS and the financial ledger, and the manual workarounds that persisted and are now informal process.
- Vendor performance against contractual service levels and the cost of the extension.
- Whether the project's own lessons-learned actions were implemented, since the next rollout will repeat the pattern otherwise.
What would make it worthless. Reviewing only the project's own success criteria; relying on the sponsor's dashboard without re-performing the calculations; starting after the team has dispersed so no one can explain the numbers; and treating go-live survival as success. A review that concludes the system is live and therefore the project succeeded has measured nothing. The valuable output is a short list of benefits not realised and controls not operating, with owners.