Oct 25, 2026Sourcing Guides
3PL Onboarding Checklist: The First 30 Days of a Fulfillment Handover
The signature is the easy part. The first 30 days decide whether stock arrives clean, orders flow and exceptions find an owner before customers do.
The signature is the easy part. A new fulfillment partner becomes real when the first carton arrives, the first order drops and the first thing goes wrong. The first 30 days decide whether those moments produce a fix or a fire.
A handover is not a welcome call. It is a set of decisions and records that must exist before the first live order. The work is unglamorous: names, fields, cut-off rules, test orders and a shared log. That work is also what keeps a good partner from looking bad in week one.
Our China fulfillment work is for stores that already ship. This is the 3PL onboarding checklist we would use in the first 30 days.
Day 0: Name One Owner and One Decision Log
We open the log before the kickoff. Name one owner on your side and one on theirs. Not a committee. A committee can approve a plan and still leave nobody responsible for a SKU that was never loaded.
Open a decision log before the kickoff. Every entry gets a date, a decision, an owner and what changes because of it. The log is not meeting minutes. It is the place where "we will use the new return address" becomes a line someone can find three weeks later.
Write a short RACI for the moving parts: SKU setup, inbound receiving, systems, returns, customer communication and escalation. Each row needs a name, not a department. If two people own the same row, nobody owns it. Put the log where both sides can edit it, and review it at the start of every call. A log that lives in one person's inbox is a diary, not a handover.
Agree how an issue moves up, with names and channels. "Email support" is not a path. A path is a person, a backup and a response expectation the partner has accepted.
Then set a freeze date. After that date, changes to the SKU master, pack rules or cut-off settings wait until after go-live. Last-minute changes are how a clean handover turns into a test of patience.
The SKU Master Is the Working Document
We freeze the master before the first inbound. The SKU master is the first document both sides should argue about. Every active SKU needs a code, a barcode, dimensions, weight, pack quantity and a flag for fragile or special handling. If the item is sold as a bundle, the bundle needs a component map. If it is sold in more than one channel, the same SKU needs a channel map.
Units of measure cause quiet errors. A supplier may quote by the carton, the warehouse may receive by the piece and the store may sell by the pack. Write the unit for each field. A picker who reads "12" without a unit can send twelve pieces or twelve cartons. Variants are the next trap. A shirt in five colors and four sizes is twenty sellable SKUs, and each one needs its own barcode and pack rule. If the master lists the parent only, the warehouse will invent the children at receiving.
Decide who maintains the master. The partner can hold it, but you own the product facts. Set a change request format and a response time. Ask how the partner handles a barcode that will not scan or a weight that is missing. The answer should be a named person and a quarantine step, not "we will fix it."
Freeze the master before the first inbound arrives. A late SKU is not a data entry problem. It is a carton that sits on a receiving floor while the store takes orders it cannot ship.
Receiving: Agree What Counts as a Clean Inbound
We agree the count method first. Receiving is where a handover shows its quality. Agree on the inbound notice, the count method and the condition check before the truck arrives.
Ask what the partner counts and what it samples. Carton count and unit count are different checks. A carton can be sealed and still be short. Decide which one applies to which inbound and who pays for the difference. Put the answer in the decision log.
Agree on the evidence. A discrepancy needs photos of the carton label, the seal, the pallet and the count sheet. A message that says "short" without a photo is an argument, not a record.
Name a quarantine area for damaged, unknown or unlabeled stock. The partner should not put an unidentified carton into pick stock because the system will accept it. Decide who can release a quarantined item and what the customer sees while it waits.
Ask for a receiving report while the inbound is still fresh. The fields matter more than the format: cartons expected, cartons received, units counted, exceptions, photos and a name. If the report arrives after the stock is already picked, the window to fix it has closed. The operational side of these checks is described at (/b2c-fulfillment).
Test Orders Before Live Orders
We go live on test results. Do not go live on a calendar date. Go live when the test orders pass.
Build a test set from real SKUs, not demo items. Include a single-item order, a multi-item order, a bundle, a backorder or preorder (if you use either), an address error, a cancellation after pick and a return. Each one exercises a different handoff.
Watch the full path. The order should reach the warehouse, then be picked, packed, labeled and shipped. The tracking number should return to the store. The inventory should decrement once, not twice. The customer email should carry the right service and the right address.
Test the exception path on purpose. Put a wrong address on one order. Cancel another after the pick list is printed. Send a return to the new address before the first real customer does. A 3PL's reaction to a test exception is a preview of its reaction to a real one. Run a return through the same test. The parcel should reach the agreed address and be opened by the named person; the unit should then be photographed and given a disposition. A return that disappears into a back room during the test will disappear in production too.
Write down expected and actual for each scenario. A test that nobody records is a demo. The sign-off should list what passed, what failed and who fixes each failure. If a failure is accepted as "known", it needs a workaround and a date.
The First Live Week: What to Watch Daily
We keep the daily call short. The first live week is not the time to disappear. It is the time for one daily call and one shared view of the same numbers.
Watch the backlog before the cut-off, not after. Watch out-of-stock items that still show as sellable. Watch wrong-item reports, label failures and tracking numbers that never return to the store. Watch the same SKU appearing in more than one exception.
Keep the meeting to fifteen minutes. Walk the same fields in the same order every day. The point is not a report; it is a decision. Each exception gets an owner and a next action.
Customer communication needs a named owner too. Decide who sends the delay note, who answers the "where is my order" message and what template they use. If the partner sends it, approve the wording before the first live order. If you send it, agree how fast the partner flags the problem.
Do not change the SOP in the middle of the week without a log entry. A quiet change breaks the test baseline and leaves the next shift guessing.
Returns and Exceptions in the First Month
We review exceptions weekly. Returns are where a new partnership often shows its real operating model. The forward flow can look fine while returns pile up in a corner.
Agree on the return address, the receiver and the condition check. Decide who opens the parcel, who photographs it and where the photos live. A return without a condition record becomes a credit argument.
Write disposition rules before the first return. Which items go back to stock, which go to quarantine, which need refurbishment and which are disposed of? Who makes the call when the item is borderline? A rule that says "inspect and decide" is not a rule.
Decide how the customer is told. A refund, a replacement or a repair is a commercial choice, but the trigger and the owner are operational. Put both in the decision log.
Lost and damaged parcels need the same treatment. Agree who files the claim, what evidence is collected and who talks to the customer. The answer may depend on the carrier, but the handover should still name one owner on your side.
Review exceptions weekly, not one by one. Look for a pattern: the same SKU, the same carrier, the same address type, the same pack station. A pattern is fixable. A pile of one-offs is just noise. Keep the return photos with the order record, not in a chat thread. The next person who handles the same customer needs the same facts.
Day 30: What "Stable" Means
We close the month with owners. Day 30 is not a celebration. It is a review against the handover, not against a feeling.
Stable means every open exception has an owner and a status. It means the inventory count reconciles to the method both sides agreed on. It means the reports arrive on the agreed cadence with the agreed fields. It means the SKU gaps and mapping issues from the first inbound are closed, not parked.
Stable does not mean zero errors. It means known errors with owners. A partner that hides a problem to keep the dashboard clean is less stable than one that reports it and fixes it.
Close the first month with a short improvement list. Pick the three changes that would remove the most recurring work. Put them in the decision log with owners. Then set the next review.
If the first 30 days are still full of unknowns, do not add volume to hide them. Fix the handover first. The second month is easier when the first month left a record.
Send the handover plan, plus the SKU and volume structure you are working with (/contact-us). We compare it with the operating checklist, mark the gaps and come back with a handover checklist and an exception responsibility table. The starting point is a live store at a steady five orders a day or more. We do not take pre-launch shops or stores still in product testing.
---
