ERP Cutover at Ridgeline Components: Go/No-Go With $2.9M Unreconciled

Recommendation: do not go live on the current date; approve a conditional cutover only after four specific gates are cleared. If the gates cannot be cleared, run the close on the legacy system even if it costs a parallel period.

The business case for the date is 'avoid a dual-system close.' That is a convenience argument, not a control argument, and it is being used to price a risk the finance function will carry for years.

The four facts that matter

1. Master data is not fit to migrate. 180,000 suspected duplicates is 15% of the item master; 9,000 unowned items means no one is accountable for replenishment or valuation. Duplicates do not just cause bad reports - they double-count inventory and split purchasing history. Migration does not cleanse data; it permanently encodes the mess and removes the legacy views that let people work around it.
2. Interfaces are unproven in the places that move money. 8 of 14 untested, and both tested failures are financial (tax determination affects invoicing; EDI pricing affects revenue). An interface that has never been run end-to-end in test has an unknown failure mode in production.
3. Three different inventory numbers. $48.0M book, $45.1M count, $46.2M from conversion test 3. The conversion logic reproduces neither source. That is not a variance to be explained at close; it is evidence the mapping is wrong.
4. The controller leaves in three weeks. The person who can explain a variance at close will be gone before the first close under the new system. Knowledge transfer must be complete and evidenced before sign-off, or she should be retained on contract through the first two closes.

Conditions that make go-live defensible

• All financially significant interfaces tested end-to-end, with signed results and a documented exception process.
• Duplicate suppression rule deployed and validated; a named owner for every remaining item above a threshold.
• Opening inventory agreed to the cycle count within a stated tolerance, reconciled item by item above a dollar threshold.
• Legacy system available read-only for at least one full close, and a documented rollback or parallel-run plan.

What would make it indefensible

Proceeding while any interface that touches tax, pricing, or inventory valuation is untested; or proceeding on a premise that the first close will 'sort out' the inventory difference.

Reject the proposed treatment. Posting the conversion output because it is 'the system's number' is a plug with better branding. Opening balances must be supportable, and the conversion test is not evidence of what Ridgeline owns.

First, decompose the difference rather than netting it. Three numbers imply more than one error, and errors in opposite directions can hide inside a $2.9M net figure:

• Book vs count ($2.9M): candidates are unrecorded scrap or obsolescence, consignment or customer-owned stock counted as Ridgeline's, receipts recorded but never physically received, unit-of-measure errors (each versus case), and standard cost that has drifted from actual.
• Count vs conversion output ($1.1M): this is the more alarming gap, because the count is physical evidence. If the conversion moved 1.1M above the physical count, the likely cause is duplicate items migrating as separate records, or quantity conversion errors on the 180,000 flagged duplicates.

Second, work by materiality slice. Rank items by extended value; reconcile every item above a set threshold (say $10,000 extended) individually, and accept statistical treatment for the tail only if the tail is demonstrably immaterial. Do not spend three weeks on 1.2 million records that contain 40,000 that matter.

Third, book the correction in the right period. Inventory adjustments belong in the period whose financials they affect - before cutover - not as an opening balance adjustment in the new system, which buries a legacy problem inside a system conversion and makes it unrepeatable and unexplainable to the auditor.

Fourth, treat the residual honestly. If a residual remains unresolved at cutover, it must be quantified, described, and signed off by the CFO as an estimate with a stated range - and disclosed to the auditor before the close, not discovered during it. An unexplained opening balance in the largest asset on the balance sheet is exactly the kind of item that becomes a comment, and possibly a deficiency, in an audit.

Pre-cutover (control of last resort before the door closes)

• Master data locked: no item creates or changes without workflow approval from the data owner, effective immediately and for 60 days post-go-live.
• Opening balance package signed by the new controller, the CFO, and the outgoing controller - three signatures, because two of the three will be gone or changed.
• Interface control totals captured from the legacy system on the final night and retained, so every post-go-live file can be proven complete.

Through the first two closes

• Reconciliation for every interface, every run: record counts and dollar control totals transmitted vs received, with an exception queue, an owner, and a 24-hour clearing SLA. Unreconciled items do not age quietly.
• Inventory valuation parallel run: one full cycle of valuation computed both in the new system and in a spreadsheet from legacy extracts, compared at the item-class level before the close is certified.
• Three-way match monitoring: purchase order, receipt, and invoice exceptions reviewed weekly rather than at month-end, since cutover errors surface as match failures.
• Journal entry review: all manual entries above a threshold reviewed by someone independent of the preparer, with specific attention to entries that move value between inventory and cost of sales.
• Audit log review: privileged and master-data changes reviewed monthly with evidence retained.
• Access recertification at 30 days, because a new system means a new role model and role drift is immediate.

How you would know it worked - success criteria stated in advance

• Zero unreconciled interface items above $50,000 at each close.
• Inventory valuation variance to the parallel run within a pre-agreed percentage.
• Close completed within the target number of business days.
• No emergency access grants outstanding at 30 days.
• Post-implementation review at 30/60/90 days against these criteria, reported to the audit committee.

Setting the criteria before go-live is the point: a post-implementation review conducted against criteria invented afterwards always concludes the project succeeded.