The Opening: Cutover Week Decides the Project

Six months of WMS project work, and the 72-hour cutover decides everything. Great code means nothing if cutover week falls apart. My worst experience: go-live morning revealed missing printer drivers — 40 RF devices across the warehouse sat idle for 3 hours, and half the day's orders slipped. This checklist exists to kill those "trivial but fatal" problems.

The Cutover Timeline

Staff working with PDAs during the cutover

T-72h: Freeze and Prepare

  • Data freeze: masters and inventory snapshots locked; legacy system stops accepting new documents
  • Print emergency paper documents: if systems go down, the floor can run manually for 4 hours
  • Notify all stakeholders: carriers, customers, 3PL partners — cutover window and emergency contacts
  • War room ready: collocated workspace, network, projector — all tested

T-48h: Final Load and Validate

  • Final data load: masters, inventory, open documents
  • SKU-by-SKU inventory reconciliation: any variance over 0.5% gets root-caused — no "we'll fix it after go-live"
  • Last integration-test round: at least 3 real business transactions per WMS↔ERP interface
  • 100% hardware check: RF devices, printers, access points, scanners — every single unit, no sampling

T-24h: Rehearse and Prepare to Roll Back

  • Full-process rehearsal: receive–putaway–pick–ship–returns, 3 orders each including one exception flow
  • Rollback drill: rollback triggers, steps, and how the legacy system resumes — documented and rehearsed once
  • Duty roster plus escalation contacts: on the wall and on everyone's phone
  • Implementation consultants confirmed on-site: no leaving on go-live day or through T+3 (in the contract)

Go-Live Day: Dual Verification and Frequent Reconciliation

  • First 50 orders dual-verified: system document plus physical goods, two signatures
  • Reconcile every 2 hours: inbound/outbound quantities, inventory balances — investigate variances immediately
  • War room collocation: every key role in one room, no overnight issues
  • Status sync to management every 4 hours: green / yellow / red, one sentence

T+24h / 48h / 72h: Close Out and Review

  • Step reconciliation down: 2 hours → 4 hours → 8 hours
  • Issue-list clearance target 90%+: the rest moves to steady-state operations tracking
  • Post-mortem at T+72h: what worked and what didn't, documented and archived

Checklist at a Glance

| Phase | Key items | Done criteria | |---|---|---| | T-72h | Data freeze, emergency docs, stakeholder notice | Zero new documents in legacy system | | T-48h | Data validation, interface testing, hardware check | Inventory variance <0.5%, 100% hardware pass | | T-24h | Full rehearsal, rollback drill, duty roster | 15 rehearsal orders, zero failures | | Go-live | Dual-verify first 50, 2-hour reconciliation | On-time order rate >98% | | T+72h | 90%+ issues cleared, post-mortem | Review document archived |

Field Case: 99.1% On-Time Rate Across 72 Hours

An e-commerce warehouse shipping 20,000 orders a day followed this checklist: T-48h caught 37 SKUs with inventory variances; T-24h rehearsal found 2 printers with driver issues — all fixed before go-live. The 72-hour cutover ran at 99.1% on-time order rate with zero rollback. The project manager's summary: "Every line on this checklist is tuition someone else already paid."

Pitfalls to Avoid

  • Never schedule cutover within 2 weeks before peak. The system needs at least two weeks of stability before its first big test — rushing cutover into peak season is gambling.
  • Define rollback triggers in writing, in advance. E.g., "4 hours unable to ship normally" triggers rollback — without written criteria, nobody dares make the call when it counts.
  • Don't let consultants leave on go-live day. Contract it: on-site support on go-live day plus T+3. It's insurance.

War Room Roles

Cutover week's war room needs five roles, no exceptions:

| Role | Responsibility | Coverage | |---|---|---| | Commander | Decisions, escalation, external updates | Full duration | | Business lead | Document verification, exception calls | Full duration | | WMS consultant | Configuration, troubleshooting | Go-live day + T+3 | | ERP interface lead | Interface monitoring, replays | From T-48h, full duration | | Hardware lead | RF devices, printers, network | From T-48h, full duration |

Two 15-minute standups daily, covering only three things: current status (red/yellow/green), blockers, and who resolves what by when. No solution debates in the standup — take details offline. Send the notes to the management group so everyone shares one picture of progress.

Writing a Rollback Plan That Actually Works

A rollback plan isn't one sentence saying "we'll roll back if we must" — it nails down three things:

  1. Quantified triggers. E.g., "ship volume below 50% of normal 4 hours after go-live," "core interface down 2 hours unrecoverable," "inventory variance over 5% unexplained." Unquantified triggers mean nobody dares make the call when it counts.
  2. Hour-by-hour steps. T+0 stop new-system documents, T+1 legacy resumes order-taking, T+2 data re-injection, T+4 full verification — each step with an owner and acceptance criteria.
  3. Rehearsal. The rollback drill runs standalone inside the T-24h rehearsal. An unrehearsed plan is no plan.

Also prepare a "degraded operations" middle state: when the system is partially up, paper documents plus after-the-fact entry buy 4 hours for troubleshooting. Many small incidents don't need rollback — they need a 4-hour buffer.

The Takeaway

The 72-hour cutover is about turning "unknowns" into "knowns": freeze, validate, rehearse, dual-verify, reconcile frequently. Follow this checklist and cutover stops being a thriller — it's a documentary: steady, methodical, done on time.