Sub-assembly sourcing: buying a module instead of its parts

Drawing the boundary where a failure can be attributed

A good module boundary is one where you can tell, from a test at goods receipt or at the point of use, whether the delivered item is good. A bad boundary hides a characteristic that only reveals itself after your own downstream operations, at which point the argument about cause becomes unresolvable. Choose divisions along functional lines with measurable outputs — a hydraulic manifold that can be pressure tested, a wiring harness with continuity and insulation checks, a gearbox with a running test — rather than along convenient physical lines that leave the important behaviour unverifiable until final product test.

The interface control document does the real work

For a single part a drawing suffices. For a module you need a document that fixes the mating dimensions, the mechanical and electrical interfaces, the environmental and functional performance to be met, the tests demonstrating it, the labelling and traceability required, and which internal choices the supplier may make freely. That last clause is what people forget. Left unstated, the supplier will optimise internal components for its own cost, which is often fine and occasionally catastrophic when it swaps a bearing, an elastomer or a semiconductor without telling anyone.

Directed components and the cost transparency they buy

Buyers often nominate specific components inside a module — a connector family, a controller, a bearing — either because approval already exists or because the value is large enough to want visibility. The supplier then buys to your direction, sometimes at your negotiated price. This preserves control and cost transparency, and it fragments accountability: when the directed component fails, the assembler will point at your choice. The workable arrangement states plainly which items are directed, who carries the risk on each, and what the assembler is still obliged to verify on receipt.

What leaves the building with the work

Outsourcing a module transfers the assembly knowledge with it. Fixtures, test rigs, work instructions, the tribal understanding of which step causes most of the defects, and eventually the people who knew it. Bringing the work back a few years later is not a reversal but a rebuild. That is not an argument against outsourcing; it is an argument for being honest about whether the module is close to what the product does or peripheral to it. Peripheral modules travel well. Modules embodying the reason customers buy from you travel badly.

Warranty, field failure and who pays for the truck roll

When a module fails in service the material cost is trivial next to the cost of getting to it. Agree in advance how field failures are allocated, what evidence triggers supplier liability, whether returned units are analysed jointly, and what happens when analysis is inconclusive — which it often is. Also settle who funds a containment action across finished stock if a batch defect emerges. Contracts silent on these points push every incident into an ad hoc negotiation conducted while a customer waits. Set the analysis route and the evidence threshold while both parties are still calm, and revisit it after the first real incident. Ambiguity here is expensive precisely because field failures arrive at the worst moment for both parties.

Frequently asked questions

How do you decide which modules to outsource and which to keep?

Weigh how close the module sits to what makes your product distinctive against how much assembly effort it absorbs. Peripheral, labour-intensive modules with clean, testable interfaces are natural candidates. Modules carrying the performance customers actually pay for should stay, even where the arithmetic favours moving them, because losing the process knowledge is irreversible in practice. Also check whether the interface can be verified on receipt; if it cannot, the module is not ready to leave.

What stops a module supplier changing internal components without telling you?

A written change-notification obligation covering the internal content, tied to specified consequences, plus an approval requirement for anything affecting a designated characteristic. Support it by keeping your own record of the internal configuration approved at part sign-off, so a change is detectable rather than a matter of the supplier's word. Periodic teardown of a delivered unit against that record is cheap and finds substitutions that paperwork never surfaces.

Does buying a module always reduce cost?

It reduces your assembly labour and overhead absorption, and adds the supplier's overhead and margin on work you were doing yourself, plus logistics on a bulkier item. The gain is real where the supplier has genuinely lower assembly cost, better volume across several customers, or automation you cannot justify. Where it merely moves the same manual work to a similar cost base, the arithmetic is thin and the added interface risk may not be worth it.