In this guide 6 sections
Buy when the work is standard. Build when your edge sits in something a package forces you to bend, such as client-specific pick rules, odd marketplace feeds or a returns grading flow. Most centres buy the core system for stock and shipping, then build the few edges that set their floor apart, keeping the join between the two clean.
Start from what makes your floor different
List every step a parcel takes in your building, from the inbound ASN to the outbound manifest, and mark each one as ordinary or particular to you. Ordinary steps suit a ready system.
Particular steps decide it. A fulfilment centre serving ten brands may run ten rule sets: kitting for one, gift notes for another, batch and expiry capture for a third. A package fakes the rest.
Be honest about volume and client mix. A single-brand operation with steady orders needs something very different from a multi-client centre that onboards a new seller every month, bringing fresh label formats each time.
- Inbound formats: ASN files, paper packing lists, pallet labels
- Storage rules: bin, shelf, batch, expiry, serial
- Per-client packing and kitting instructions
- Carrier mix and label formats
- Client reporting and invoicing terms
Test marketplace integrations first
Marketplace and storefront feeds are where fulfilment systems differ most, so ask how orders arrive, how stock levels go back and how cancellations and tracking numbers return. Then ask about rate limits, and about the retry behaviour when a feed fails at midnight while the night shift is packing against stale stock levels.
A maintained connector for your marketplaces saves months of work, while one that sits on a feature list but breaks after a changed API is worse than none. Phone a reference customer who sells on the same marketplaces as you, and skip the polished demo account that hides every failed sync.
Selling on many small channels? Then a custom order layer can wrap them all and absorb each API change in one place instead of patching five connectors each time a channel changes.
Look hard at pick paths and returns
Pick logic drives throughput. Check whether the system supports wave, batch and zone picking, cart or tote assignment, and short-pick handling, then ask how it re-sequences bins after the layout changes.
Returns are the other test. A returned item must be scanned, graded, then restocked or routed to refurb, and many warehouse packages treat all of that as an inbound receipt with a flag. A fulfilment centre needs grades, photos, reason codes, a decision per unit and a credit note raised to the client before the stock ever returns to a bin.
Heavy returns justify a custom module. Keep the bought core beneath it.
- Wave, batch and zone picking, with short-pick handling
- Bin sequencing after re-slotting
- Scan, grade and disposition steps for returns
- Photo capture and reason codes
- Credit notes back to the client
Count the costs that are not on the quote
A bought system brings licence fees, per-order charges and upgrades you do not control, while a custom build brings design time, testing, hosting and a maintenance duty. Neither is free of surprises.
Find out who holds the data model and the code, how you export your stock ledger if you leave, and what happens when your lead developer goes. Those answers outweigh the first quote.
Scanners and printers bite. Handheld compatibility and label formats often constrain the choice more than any software feature, so put your thermal label printers and rugged scanners on the trial list.
Choose a path and test it on real orders
The middle path usually wins. Buy a core for inventory, locations and shipping, then build the pieces that carry your advantage: client portals, a returns flow or a rules layer for marketplace orders. Guard the seam between them.
Before signing anything, run a day of your real orders through a trial that includes bad moments: a cancelled wave, a damaged inbound pallet and a returned kit. No brochure admits what a trial exposes.
Write down the three flows you cannot bend, then ask each vendor to demonstrate those before showing a single dashboard or report screen.
Build or buy scoring sheet
One sheet lists inbound, put away, pick, pack, ship, returns, marketplace feeds and client reporting as rows, with columns for how standard each step is. Further columns note what a build would take, who owns the risk and the final call: buy, build or bridge, plus what a ready system offers there. A footer lists three trial flows.
Open a blank worksheet to printQuestions owners ask
When is a ready warehouse system clearly enough?
When your flows match standard receive, pick, pack and ship patterns, you serve few clients and your marketplaces have well-maintained connectors for each channel you sell on. Customising then adds cost, not advantage.
Is a custom build only for large centres?
No, because size matters less than how odd your client rules are, and a small centre with unusual requirements can gain more than a large one doing uniform work.
Can we start with a ready system and build later?
Yes, provided the system exposes its stock ledger and order history through an API or scheduled exports, so a later module can read the same records. Check that early. A closed system makes the move harder.
Who should be involved in the choice?
The floor supervisor, the inventory controller, the person who handles client onboarding and whoever manages marketplace accounts, because each of them sees a different weak point.