Switching your 3PL without stopping shipments

A 3PL switch has one hard constraint: the shipping does not stop while you do it. Everything else in the plan bends around that. The work is not choosing the new partner, it is sequencing the handover so that at every hour of the move, some warehouse can pick your orders.

Reconcile the stock before a single carton moves

Your system says one number. The old 3PL’s WMS says another. The physical count says a third. All three are wrong in predictable ways: units picked but never manifested, customer returns sitting in a quarantine bin nobody has processed, damages the 3PL wrote off in their system and never passed to yours, and short-shipped inbounds from months ago that were never closed.

Do a joint physical count with the outgoing partner present, SKU by SKU and batch by batch, with the unsellable pile counted separately. Sign it. That signed number does two jobs: it becomes the opening balance at the new site and the closing claim against the old one. Move stock before it is signed and you have no claim, because every shortage after that point is arguable.

Budget more time than feels reasonable. On a few thousand SKUs this is days, not hours, and it usually has to happen alongside normal dispatch.

Move by SKU class or by channel, never all at once

Two sequencing patterns work. Move slow movers first, because the old warehouse is happy to release them and a stockout on a tail SKU costs very little while the new site learns your pick paths and packing rules. Or move by channel: your own D2C first, marketplace self-ship second, platform-fulfilled inventory last, because that last one carries inbound appointments and its own scorecard.

What does not work is a single weekend truck with everything on it. It gives you one shot at a clean receive, and a receiving error at the new site stays invisible until orders start failing.

The stock already promised to open orders

At the moment the truck loads, some units in the old warehouse are allocated to orders that are accepted and not yet shipped. You have two choices and both cost something.

Let the old node ship its open book to zero before loading. That is cleanest, and it adds days to the window where nothing else moves. Or de-allocate, re-promise from the new node, and tell customers and channel teams the dispatch date has shifted. That is faster, and you pay in dispatch SLA and late-shipment metrics, which is a price worth quoting back to whoever set the timeline.

Whichever you pick, freeze allocation at the old node at a stated time and publish that timestamp to support, to the channel teams and to the incoming 3PL.

The dead window

There is always a stretch where the old site has stopped picking and the new one is not live. Assume it exists and shrink it deliberately.

• Load item masters, packaging rules and channel mappings into the new WMS a week early and test with a dummy order end to end.
• Move in waves so the dead window applies to a slice of the catalogue rather than all of it.
• Schedule the physical move into your lowest volume days, away from any platform sale event or purchase order cycle.
• Pre-position a small buffer of your top ten SKUs at the new site before the main move, so day one has something to ship.

Couriers, rates and the return address

If you have been shipping on the 3PL’s courier contract, those rates belong to them and they do not travel with you. Find out before you sign the new agreement, because the cost per order you have been quoting internally may not be reproducible at the new partner.

If you ship on your own aggregator or direct courier accounts, the account moves but three things change: the pickup location, the pincode serviceability profile from a new origin, and the return address on every label. Serviceability matters more than people expect, because the delivery promise you display on your own site was calculated from the old origin and nobody remembers to recalculate it.

Returns to an address you no longer occupy

On the day you leave, there are RTO shipments and customer returns already inside the courier network addressed to the old warehouse. They keep arriving for weeks.

Agree in writing, before you move, that the old partner receives and holds them for at least one full return cycle at a named per-unit handling rate, and forwards them in consolidated batches. Without that agreement they get refused at the gate, marked undelivered and eventually disposed of, and you carry the loss twice: the unit and the refund. Update the return address at the courier and on the channel side on the same day, and accept that labels already printed cannot be changed.

Get your data and your bad stock out

Ask for batch and expiry data, serial numbers where you carry them, the damages and shortages log, and the photographic evidence attached to past claims. Ask for it in a loadable format, and ask while the relationship is still functional rather than after the last invoice is disputed.

Then deal with the unsellable pile you counted separately. Damaged, expired and returned-unsellable stock has to be moved, destroyed against a certificate, or liquidated, and the outgoing partner wants it gone faster than you want to decide. Decide before the final count, because after the count it becomes a line item somebody is charging you storage on.