Bottom line first: in multichannel fulfillment order batching, get the wave rules right and your warehouse saves a full shift of rework every day. Get them wrong, and no amount of fast picking makes up for it. I've led 30-plus go-lives; wave-rule problems almost always show up on the first big promotion night.
What is a wave? An OMS groups orders by a rule — channel, service promise, location, carrier cutoff — and releases them to the WMS for picking. The hard part is when the four dimensions fight.
In November 2025, a home-goods brand in Shenzhen hit this wall during Black Friday week. Their OMS had one wave rule: first come, first served, a wave every 120 orders. Amazon orders had to ship the same day, the website promised three-day delivery, and the TikTok shop's cutoff ran two hours earlier. Picking finished before the pack bench discovered a batch of Amazon orders would miss the day's carrier cutoff. They were pulled out and rushed separately; the team worked four extra hours and sent two extra expedited runs. The post-mortem found it: the wave rule had no "service promise" dimension. I've seen this pitfall three times — every time the team says the same thing: "We thought first come, first served was the fairest."
Fairness is not the goal of a wave rule. On-time is.
Why Multichannel Fulfillment Order Batching Goes Wrong
Level one is split shipments. One order picked across two waves, or two warehouses, means two parcels. On the Amazon MCF side, fulfillment fees are charged per item, by size tier and speed — starting at $7.15 per item, with tiered discounts for multi-unit orders; check Amazon's current rate card first. Split one order into two parcels and the discount is gone. In your own warehouse, a split is one more pack, label, and verification.
Level two is rework after the cutoff. The wave is released, picking is halfway done, and someone realizes a group of orders has an earlier cutoff than expected. Two options: pull those orders out, or replan the whole wave. Pulling orders means changing pick tasks, pack sequence, and labels; replanning wastes the last half hour.
Level three is last-mile surcharges. The wave is grouped by location — but the destinations are scattered across five or six carriers and seven or eight service products. The pack bench keeps switching label templates and consumables, and the freight bill picks up small-batch surcharges. The 20 minutes picking saved, packing pays back double.
Multichannel Fulfillment Order Batching: The Configuration Checklist
This is the one page in my go-live night checklist that never moves.
Step one, set the layering order — don't lay the four dimensions flat. My order: service promise first, channel second, location third, carrier cutoff as a hard boundary. Split orders by promised service into "ship today," "ship tomorrow," and "ship in three days." Within one tier, split by channel — packing and label formats differ by channel. Within a channel, cluster by location. Then lock it down with the cutoff: a wave must be released before its earliest cutoff, with buffer for picking plus packing.
Step two, cap wave size — by pick lines and volume, not order count. One wave of 120 orders at 2.3 lines each is 276 lines — 4.6 labor hours at 60 lines per picker per hour. Overshoot the cap and the wave slips into the next cutoff window — so write the cap into the OMS.
Step three, build an exception lane. Keep an "expedite insert" lane: high-priority orders, VIP customers, and already-delayed orders skip the normal queue into an expedite wave. Keep the lane under 5% of total volume, or it hollows out the normal waves.
Step four, lock inventory before releasing the wave. Before any wave goes out, the OMS must confirm every order in it has locked, allocated stock. Release a wave without it and the whole wave stops for replenishment when a shortage shows up mid-pick. I've seen this pitfall three times — always in the first week after go-live.
Step five, write down how waves hand off to each other. The previous wave isn't finished — do you release the next one? Look at the cutoff: if it has arrived, release it; hold the unfinished part of the previous wave and push it to the front of the next wave's queue. Never let one wave block the whole line.
| Group by | What you save | What you pay | Best for | | Service promise | On-time rate | Longer paths | Multichannel, big service gaps | | Channel | Standardized packing | Locations walked twice | Many channels, different packing | | Location | Shortest walks | Cutoffs get ignored | Single channel, concentrated SKUs | | Carrier cutoff | Most reliable at cutoffs | Fragmented waves | Many cutoffs, dense departures |
No standard answer: service promise and cutoff are hard rules — violating one raises an alert; channel and location are soft rules — violating one writes a log entry.
Correcting a Bad Wave
Even good rules produce a bad wave sometimes. Write the correction workflow in advance; don't wait for the incident.
First, freeze: in the OMS, change the wave to "frozen" and move what's already picked to a holding area, not the pack bench. Freeze fast, within five minutes, or pickers keep working the wrong wave.
Second, triage: split the orders into "can still make the cutoff" and "already missed." The first group gets re-batched into a recovery wave, picked first. The second gets a revised promise and a call to customer service — don't let the customer find out from tracking first.
Third, review — three questions: which rule was misconfigured? Why didn't testing catch it? What check will catch it next time? Write it back: change one rule, add one test case.
In September 2024, a 3PL warehouse in Los Angeles gave me a textbook case. Their wave rule grouped by location, and pick productivity was genuinely high. But for three weeks running, every Friday afternoon a batch of orders missed the cutoff. Two weeks of digging found it: Friday capacity was set to the midweek average, but Friday volume ran 40% higher, so waves went out late. The fix: vary capacity by day of week — release Friday waves one hour earlier and cut capacity by 15%. After that, Friday never missed a cutoff again. Wave rules aren't set once and forgotten; when the volume pattern changes, the rules change with it.
Go-Live Night Checklist: Wave Rules
On go-live night, I gather the team and walk this list:
- Is the cutoff table current — every channel, carrier, and service product checked against the contract?
- Is the wave capacity cap calculated on lines and volume — not just order count?
- Is the expedite insert lane open, capacity under 5%?
- Is inventory locked before the wave is released? In the test environment, insert a short order on purpose and see if the system blocks it.
- Has the wave-freeze and correction workflow been run end to end in the test environment? A workflow never rehearsed doesn't exist.
- Do wave-rule changes go through change approval — who changed it, why, when it takes effect, can the OMS show that?
Six ticks, then sleep.
Here's the counterintuitive part: a fuller wave is not a more efficient wave. Many warehouses push the OMS to pack every wave full. The result: waves go out late, cutoff windows shrink, error rates rise, and rework eats everything picking saved. I do the opposite: leave 10% of wave capacity empty for high-priority orders arriving before the cutoff. The empty space looks like waste; it's breathing room. The goal of a wave rule is not to fill the wave. It's to be on time.
The system view and the seller view look at two sides of the same bill: the seller cares which channel fulfills an order cheapest; the system cares how to batch orders so nothing splits, nothing gets reworked, no surcharge tier gets tripped. For the channel side of MCF, read our Amazon Multi-Channel Fulfillment (MCF) guide. That piece covers the MCF channel side; this one covers wave configuration inside your own OMS.
Let me close with the math — an illustrative example, not real warehouse data: 1,200 orders a day, misconfigured rules splitting 8% of orders, at $3.50 extra per split order, over 22 working days a month — that's 1,200 × 8% × $3.50 × 22, about $7,392 a month. Fix the rules, push the split rate to 2%, and you save over $5,500 a month. Swap in your own ratios. But the formula is real: every point off the split rate is money.
Review the wave rules every quarter — channels get added, cutoffs change, volume patterns shift. Configure and forget them, and in six months they're wrong.
By Lin Xiao, WMS Implementation Consultant








