Ask five warehouses what "kitting" means and you get five answers. That confusion costs money, because the two main kinds of kitting need opposite workflows, and running the wrong one is how bundles oversell and components vanish. Here is the distinction, and how to decide which one your operation needs.
The two kinds of kitting
Build-to-stock (assemblies). You make the kits ahead of time. A work order says "build 50 gift boxes," someone picks the candles and the crinkle paper, assembles at a bench, and 50 finished gift boxes go into a bin as real inventory. When an order arrives, the picker grabs a finished box like any other SKU.
Kit-on-demand (virtual kits). Nothing is pre-built. The bundle exists only as a bill of materials. When an order for the bundle arrives, the system explodes it into component picks, and the packer assembles it at the packing station, once, for that order.
When build-to-stock wins
- Assembly takes real time or skill (sealing, labeling, quality checks). Doing it in a batch of 50 at a bench beats doing it 50 times under shipping pressure.
- Demand is predictable. Pre-building smooths the workload into quiet hours.
- The kit ships in its own packaging that has to be prepared anyway.
The tradeoff: capital sits in finished kits, and if the promotion flops you must break kits back down. A WMS that supports disassembly (finished kits back into components, counts corrected automatically) removes most of that risk.
One more thing good software should do here: notice the shortfall before you do. If bundle orders are outpacing finished kits, the system should flag it and propose the build, with quantities worked out from component stock, rather than waiting for a picker to find an empty bin.
When kit-on-demand wins
- Components sell individually AND in bundles. Pre-building traps shared components inside kits while the standalone listing sells out.
- Bundles change often. A new bundle is just a new bill of materials, not a build project.
- Space is tight. Virtual kits occupy zero shelf.
The tradeoff: packers assemble under time pressure, so the pack screen has to make it obvious which components form which kit. If your software just shows a flat list of loose parts, mispacks follow.
The math that has to be right either way
Kit availability must be computed, not stored. A kit-on-demand bundle's honest stock number is the minimum across its components: if the bundle needs 2 of A and 1 of B, and you have 100 A and 38 B, you can sell 38 bundles, not 100. Systems that let you type a stock number for a virtual kit are how overselling happens.
Components must decrement exactly once. Build-to-stock consumes components when the work order completes; kit-on-demand consumes them when the order ships. A system that decrements at both moments, or at neither, drifts within weeks.
Provenance matters. If a work order gets cancelled mid-build, someone is holding a tote of picked components, and they do not remember which bins those came from. The software should: put-back instructions belong on the screen, bin by bin.
Questions to ask any WMS vendor
- Can the same product catalog hold both kinds of kitting at once?
- Is kit availability computed from components in real time?
- What happens when a work order is cancelled after picking?
- Can you disassemble finished kits back to components?
- Do lot numbers flow from components onto builds, for recalls?
For what it is worth, OpsBox AI supports both modes on the Growth plan: bills of materials, work orders with component picking and scrap tracking, kit-on-demand orders that explode at pick and group back together at pack, computed buildable quantities, and one-click disassembly. Pricing is public and the demo takes 15 minutes.