Should kitting happen on-demand at order time or be pre-assembled and held as its own SKU in 3PL inventory?
Assemble-to-order kitting is the right default for a Shopify brand whose kit mix still changes, and pre-assembled kit SKUs win once a kit's demand is stable and forecastable. Pre-kitting cuts the per-order pick touch and locks components you can no longer sell separately. Shopify Bundles caps fixed bundles at 30 components and tracks stock overall, never by location (Shopify Help Center, September 2026).
The two costs you are trading
Assemble-to-order pays labor on every order and keeps every component free to sell. Pre-kitting pays labor once in a batch and freezes components inside finished goods. That is the entire trade, and it resolves into numbers you can estimate before you commit.
| Criterion | Assemble-to-order | Pre-assembled kit SKU |
|---|---|---|
| Labor timing | Every order, at the pick face | Once, in a scheduled batch |
| Unit labor cost | Higher per kit; kitting labor benchmarks at $35 to $60 an hour, average $39 to $43 | Lower per kit; batch assembly amortizes setup |
| Additional-item pick fee | Applies per component; benchmark $0.30 to $0.75 per item, average $0.48 | One pick for the whole kit |
| Component flexibility | Full. Every unit remains sellable on its own | None. Committed units are gone until someone breaks the kit down |
| Inventory accuracy risk | Race conditions when two kit orders reserve the same component | Clean kit counts; the risk moves to forecast error |
| Storage | No extra footprint | Extra finished-goods footprint; pallet storage benchmarks $15 to $40 a month, average $20.17 |
| Speed at peak | Slower per order; assembly is in the critical path | Faster; peak is just picking |
| Failure mode | Cycle time and mispicks under load | Dead components inside kits that did not sell |
Benchmark figures from The Fulfillment Advisor's 2025 costs and pricing survey of 600-plus warehouses, via Fulfill.com, cross-checked June 2026. They are industry medians, not any single provider's rate card, and your 3PL quotes kitting per client.
The thresholds that decide it
- Forecast stability. Pre-kitting is a bet on demand. If you can forecast a kit's next 8 weeks within a band you would defend to your CFO, batch it. If the kit is new, seasonal or promotional, assemble to order.
- Component overlap. Count how many of the kit's components also sell on their own and how fast they move. High overlap plus fast movement is the strongest argument against pre-kitting, because every committed unit is a unit your bestseller cannot use.
- Assembly complexity. Two items into a poly bag is a pick-face task. Nine items, an insert card and a branded box is a batch task, and doing it at order time will show up as cycle time in your peak week.
- Peak concentration. If a single week carries a disproportionate share of kit volume, pre-kit for that week specifically and revert. This is the most common right answer and the least commonly chosen one, because it needs a decision date and an owner.
- Order volume. Our kitting analysis is written against 800 to 2,500 orders a day with 15 to 60 kit SKUs sharing components (Deploi reference scenario, September 2026). Below that, the labor arbitrage on pre-kitting is small enough that flexibility usually wins.
What Shopify can and cannot model
Neither mode is represented in Shopify. Native bundles decrement component stock inside the order and do not track inventory by location (per Shopify's Help Center, September 2026), which means a pooled count across nodes and no view of whether the components for tomorrow's batch are in the right building. Pre-kitted SKUs, meanwhile, look to Shopify like an ordinary product, which is convenient until you need to know how many of that product are theoretically available if you broke kits down.
The thin layer worth building is the same in both directions: component reservation so two kit orders cannot claim the same unit, a build sheet on the pick ticket, and location-aware counts. Our kitting verdict puts that at an estimated $15,000 to $40,000 over 5 to 9 weeks, longer where a 3PL's system has to accept the build sheet (Deploi estimate, illustrative, September 2026).
When NOT to pre-kit, even when the math says yes
- When the components are perishable or dated. A kit built in August against November demand ages its own contents, and the write-off is invisible until someone opens a box.
- When the kit's contents are still being argued about. Every pre-built unit is a decision you cannot revisit without paying to break it down.
- When you have no process for breaking kits down. This is the question nobody asks and the one that decides whether pre-kitting is reversible. If the answer is "we would have to call them," the commitment is permanent in practice.
- When you are pre-kitting to hit a shipping SLA you have not measured. Assembly is rarely the binding constraint on cutoff time. Carrier pickup usually is.
The Deploi point of view
Our own position, from building on Shopify. Separate from the facts above.
- Our take: Default to assemble-to-order and earn your way into pre-kitting with a forecast. The asymmetry is the argument: an assemble-to-order mistake costs you a few minutes of labor per order, and a pre-kitting mistake costs you components locked inside product that did not sell. One is a running cost, the other is a write-off.
- What we’ve seen: In five bundle builds, the reconciliation problem was worse under pre-kitting every time, because the kit SKU and the component counts drift independently and nothing forces them back together. Assemble-to-order keeps one number honest by construction.
- Times we’ve shipped this: 5 builds delivered.
- What it takes: roughly 240 hours of scoped work for fixed bundles and kit merchandising (directional Deploi estimate from a small sample of engagements, not a measured average).
- Where we disagree: 3PL sales conversations frame this as a unit-cost optimization, and the spreadsheet always favors pre-kitting because the spreadsheet has no cell for committed inventory. Add that cell before you decide. Concurrency is the other thing the pitch skips: two kit orders reserving the same component is a race condition, not a UI problem, and it is the actual hard part of assemble-to-order.
- What this page adds: the benchmark cost lines on both sides of the trade, the five thresholds that pick a mode, and the four conditions under which pre-kitting is the wrong call even when the unit math favors it.
Reviewed by Martin Dejnicki, Director of SEO & AI Search. Facts verified 2026-09-13.
Where we worked this out
Our decision records