A concrete example
A customer orders three items in one order:
- A 300g mug — lives in Warehouse Milton Keynes.
- A 2kg ceramic plant pot — Warehouse Birmingham.
- A perfume bottle — Warehouse Milton Keynes but flagged hazmat.
That order has to ship as three parcels: the mug alone (small parcel, Royal Mail), the plant pot alone (large parcel, Evri or DPD), and the perfume separately (hazmat, restricted courier). Auto-split does this assignment automatically before the order reaches the picker.
Why “just pack it in one big box” doesn't work
Five reasons, in roughly descending order of pain:
- Multi-warehouse.Items physically sitting in different warehouses can't magically teleport into one box. They have to split.
- Hazmat and restricted items.Perfume, lithium batteries, aerosols, magnets, knives — can't share a box with arbitrary items. Different courier network required.
- Courier weight and size caps. Royal Mail Tracked 24 caps at 20kg; a single heavy item plus a normal one can push a parcel over. Evri drops price dramatically below 2kg; splitting a 2.2kg order into 1.5kg + 0.7kg can save 35p per order.
- Perishable or time-sensitive items.Food items, flowers, and live plants can't wait for a slower item to dispatch. Split so perishables go out today.
- Age-restricted items.Alcohol, tobacco, and CBD need signature-on-delivery — the rest of the order doesn't need to pay that premium.
The rules engine: what a good auto-split looks like
Check for these in any platform you evaluate:
- Warehouse-level split.Table stakes. If items don't share a warehouse, the order splits.
- Attribute-based split. Flag SKUs as hazmat, perishable, oversized, age-restricted — and define rules that trigger a split when those flags conflict.
- Weight / dimension rules.“If combined weight > 10kg, split.” “If longest dimension > 1.2m, split.”
- Per-courier rules.“If item has Royal Mail Tracked requirement but another item in the order requires DPD Next Day, split.”
- Manual override in the dispatch queue. For the inevitable edge cases.
- Child-order rate shopping.Each split sub-order should pick its cheapest courier independently, not inherit the parent's.
Platform support
| Platform | Auto-split | Rule types | Child-order rate shopping |
|---|---|---|---|
| MaxInventThat's us | Yes | Warehouse + attribute + weight + courier | Yes |
| Linnworks | Yes | Warehouse + courier (custom rules) | Tier-dependent |
| Brightpearl | Yes | Warehouse + attribute | Yes |
| Cin7 Omni | Yes | Warehouse + basic rules | Partial |
| Veeqo | Partial | Warehouse only | No |
| Sellbrite | No | — | — |
| Zoho Inventory | No | — | — |
| Shopify inventory | No | — | — |
The numbers, if you care about numbers
For a seller doing 1,000 orders/month with roughly 25% multi-item orders, roughly 40% of those multi-item orders will need splitting (based on data across MaxInvent customers, Q1 2026). That's about 100 orders/month where manual splitting costs you ~2 minutes of staff time — or ~3.3 hours a month. At £15/hour loaded labour cost, that's about £50/month gone. Auto-split pays for itself fast.
More importantly: manual splitting is where dispatch errors come from. The picker sends the wrong SKU, the label is for the other shipment, returns happen. Each return costs ~£8-15 in lost margin and handling. Auto-split eliminates the class of error entirely.
How to pilot auto-split before committing
- Flag 10-20 of your most commonly-splittable SKUs with the relevant attributes (hazmat, warehouse, weight).
- Run the platform's auto-split in “suggest” mode for a week — it proposes splits but doesn't execute. Review every proposal to make sure the logic matches your packing reality.
- When confident, flip to “execute” for those SKUs only. Monitor for 2 weeks.
- Roll out to all SKUs. Keep the dispatch queue override visible so edge cases can be manually un-split.