Here is my verdict up front: a wms go live cutover plan is not a document. It is a list of times, names, and tripwires, agreed before Friday night, that tells a tired crew exactly what to do at 3 a.m. on Saturday. If your plan cannot do that, it is a slide deck, not a plan.

I have watched too many go-lives fail in the same place. Not in the software. In the gap between "the system is up" and "the warehouse can ship." Those are two different things, and a cutover weekend is where teams learn the difference the expensive way.

The go-live Saturday I will never forget: 3,000 orders stuck by 7 a.m.

October 2022, a 60,000-square-foot 3PL outside Ontario, California, shipping home goods at roughly 7,000 orders a day. Cutover started Friday at 9 p.m. Old system frozen, inventory imported, and by 2 a.m. someone posted "cutover complete" in the group chat. Everyone went home.

By 7 a.m. Saturday, pickers were standing in empty locations. More than 3,000 orders pointed at bins with nothing in them. The system was fine. The data was wrong. The pre-cutover count had covered only the fast-moving A-items. B- and C-items were imported straight from the old system's book stock, and book stock disagreed with reality in more than 800 locations. Not by a little. By enough to jam a whole Saturday morning shift.

The fix was ugly. We stopped the pick line, put 20-plus people on paper lists, and verified locations one by one, correcting the system as we went. The backlog cleared around 4 p.m. Saturday. Nearly 2,000 orders shipped a day late, and two big customers missed their delivery windows. The customers did not blame the software. They blamed us, and they were right: a problem we could have caught Friday night was discovered by pickers on Saturday morning instead.

There are two lessons here. One is obvious: do not skip the count. The other is the one that matters more. The plan had no "first real orders" checkpoint. That 2 a.m. "cutover complete" verified that people could log in and interfaces responded. It never verified that a single real order could travel from receipt to pack to ship. A system being up is not a warehouse working.

The 48-hour wms go live cutover plan I actually use

My rhythm is simple: freeze Friday night, run orders Saturday, reconcile Sunday. You can shift the clock. You cannot shuffle the order.

Friday, 8 p.m.: last normal outbound cutoff. From 8 to 9 p.m., clear the floor. Half-picked orders get finished or released back to available stock. Nothing half-done crosses into the weekend. At 9 p.m. sharp, freeze the old system. No more orders, no more location moves. Then import three things: on-hand inventory, open orders, and unfinished tasks. Before anyone declares success, reconcile three counts: total inventory, open order count, unfinished task count. If the numbers do not match, take the extra hour. Entering Saturday with wrong numbers is how you buy a Sunday of rework.

Overnight Friday into Saturday is for the switch and integration checks: orders flowing in from the OMS, carrier labels printing, scanners, printers, and scales each tested in turn. The classic mistake here is testing whether things "connect" instead of whether they "work." My rule is blunt: before dawn, run 20 real orders end to end, from order receipt through pick, pack, and ship. Twenty orders through, then you may say "cutover complete." Not before.

Saturday daytime runs at half speed. Do not go straight to full volume. Pickers are learning a new screen, and half speed is the space you give them to make mistakes cheaply. At noon Saturday, hold a 15-minute review with three numbers on the table: pick accuracy, orders per hour, and exception count. Those three numbers decide whether the afternoon speeds up or stays at half pace.

Saturday night into Sunday shifts to reconciliation: inventory variances, unshipped orders, tasks stuck in exception. By 4 p.m. Sunday, you must be able to answer one question honestly: can we open normally on Monday? If yes, staff normally. If no, tell customers and carriers before Sunday evening and move Monday pickup times. The worst move is pretending everything is fine on Sunday night and letting drivers queue at the gate on Monday morning.

Rollback in your wms go live cutover plan: who calls it, and how fast you get back

Here is the counterintuitive call I will stand behind: rolling back is not failure. A rollback is the most valuable part of a cutover plan. A cutover with no rollback plan is not brave. It is a bet, and if you lose it, the whole warehouse pays for it overnight.

Write down three tripwires before Friday, because that is what a wms go live cutover plan is for: deciding in advance, while everyone is calm. If inventory variance tops 5% of locations, or 20 picks in a row hit exceptions, stop the line and verify. Do not keep picking through errors. If by noon Saturday pick productivity is under half of normal and nobody can say why, start the rollback assessment. If orders stop flowing from the OMS or carrier labels stop printing, and the break is not fixed within one hour, roll back. Do not wait.

Name one person who can call it, and write the name down. The worst scene I have seen was 3 a.m., a real problem on the floor, five people looking at each other, and nobody willing to say "roll back." By the time someone decided at dawn, the cost of going back had tripled. My rule: one decision-maker for the weekend, usually the warehouse manager, not the IT lead. IT judges whether the technology can be fixed. The warehouse manager judges whether the business can afford to wait. If they disagree, the warehouse manager wins. Technically fixable and commercially affordable are not the same thing.

Rehearse the rollback, too. At least once before cutover, walk the full path back: how the old system comes back, how orders created after the freeze get moved back, what happens to freight already picked. A rehearsal takes about two hours. A real rollback without one takes at least six extra hours. In March 2024, a 25,000-square-foot e-commerce warehouse near Dallas rehearsed twice, including one full rollback drill. On Saturday at 10 a.m., the carrier interface genuinely went down. After 40 minutes with no fix, the team was back on the old system in 25 minutes. Only about 400 orders shipped late that day. The owner later told me those two rehearsals were the best money he spent that year.

Now the math, because this is a trade-off you should price before you feel it. Take the Ontario volume: about 600 orders per hour. Push through bad data for four hours and you create rework on at least 2,400 orders, which costs you a full extra shift on Monday. Roll back instead and you lose two hours, about 1,200 orders of capacity, but the data stays clean and Monday opens normally. A rollback costs you time. Pushing through costs you time plus trust. Customers will forgive a one-day delay. They will not forgive shipping the wrong goods.

Situation Roll back Push through
Cost 2 hours down, 1,200 orders of capacity 2,400 orders of rework
Data Back on the old system, books match reality Errors compound all weekend
Monday Opens normally Needs an extra shift of rework
Customers One day late, explainable Wrong shipments, hard to explain

One more thing most teams miss: the extra hands you need on cutover weekend are not in IT. They are in customer service. When the phone rings on Saturday morning, service needs to say "your order is one day late, it ships at this time," not "the system is changing, we don't know." Give them a one-page sheet before Friday close: which orders are affected, the expected recovery time, and who confirms it. Write it Friday. Do not improvise it while the phones are ringing.

So here is the question I leave you with. Before you set a cutover date, ask yourself: if something breaks at 7 a.m. on Saturday, does your wms go live cutover plan give the floor a sheet of paper that tells them what to do without calling anyone for permission? If not, do not set the date yet.

By Lao Mi, Supply Chain Veteran (20 yrs)