Key takeaways
- The sourcing boundary is a decision about interfaces, not only about piece price.
- A sub-assembly keeps control of the final build but adds handling, inventory and test steps in your own plant.
- A full assembly can lower landed cost and simplify the line, but it moves interface risk to the supplier.
- Choose the model per product family, and revisit it when volume, variants or freight change.
The two models, and the line between them
Sub-assembly sourcing means a supplier builds a defined group of parts, such as a contact set on a coil former or a cable with a terminal and a seal, and you complete the final product. Full assembly sourcing means the supplier builds and tests the finished unit, and your plant receives something that can ship.
The boundary is rarely a single line. A product can have three or four tiers: components, sub-assemblies, modules and the finished unit. Each tier adds a test, a pack and a set of documents. The sourcing question is which tier should sit with the supplier and which should stay in your plant.
The answer depends on where the difficult interfaces are. If the critical interface is between a stamped contact and a moulded housing, and that interface decides the electrical performance, it may be better to keep it close. If the interface is well defined and easy to test, a supplier can own it with less risk.
Control and interface risk
Every boundary you move outward transfers a set of decisions. The supplier chooses the sequence, the fixture, the operator, the test and the containment when something goes wrong. That can be a benefit, because a specialist supplier may run the process better than a general plant, but it also means you no longer see the detail unless the contract and the records give you visibility.
A sub-assembly keeps the final integration in your hands. You see the parts arrive, you control the final test and you can change the top level design without renegotiating the whole build. The cost is that you carry the interfaces between sub-assemblies: the tolerance stack, the handling damage and the inventory that buffers two schedules.
A full assembly moves that integration to the supplier. The supplier owns the stack-up and the final test, and you receive a unit with a single part number. The risk is that a failure at your customer is now a failure of a unit you did not build, and the root cause may sit two tiers down. Contractual traceability and a clear change process matter more in this model, not less.
Whichever model you choose, keep the drawing for the interface. A boundary without a controlled drawing is not a boundary, it is an assumption.
Where the cost actually moves
The piece price comparison is the easy part and often the misleading part. A sub-assembly quote looks lower because it excludes the final build, the final test and the packaging. Add those back at your own labour rate before comparing.
Full assembly usually wins on labour productivity because the supplier runs the build every day and has fixtures amortised over several customers. It can also win on freight, because one assembled unit ships in less volume than a set of loose parts with individual protection. It can lose on duty if the tariff classification changes when parts become an assembly, and it can lose on cash if the supplier needs a longer payment term to carry the inventory.
Sub-assembly can win when your plant has spare capacity, when the final configuration is customer specific, or when the last step is a test that needs your own equipment. It can also protect a core competence, such as a calibration or a software pairing step that you do not want to teach a supplier.
The honest comparison is total cost per good unit at the dock, including freight, duty, inventory, handling, test and the cost of the quality risk. A model that only compares piece prices will usually favour the option that hides the most work.
Count the interfaces, not just the parts
Each boundary adds a drawing, an inspection, a pack and a conversation. A design with fewer boundaries is often cheaper to source even when the piece price is higher.
Logistics, inventory and change speed
Loose parts are dense and easy to pack, but they need protection at the interface features. Assemblies are bulkier and more fragile at the cable or the connector, and a damaged unit is usually scrap rather than rework. The packaging cost can decide the model as much as the labour cost does.
Inventory is the hidden line. Sub-assemblies let you hold a small buffer of common parts and configure at the end, which reduces finished goods risk. Full assemblies push the buffer to the supplier, but you may still need safety stock to cover the longer replenishment lead time. Either way, someone pays for the stock.
Change speed is the strongest argument for keeping the final step close. A customer specific label, a firmware pairing or a last minute variant is easier to control in your own plant. If the change affects a supplier built assembly, every change becomes a tooling, documentation and first article exercise.
Freight mode changes the maths too. Air freight on a full assembly moves a lot of air. Sea freight on loose parts can be cheaper but adds transit time and inventory. Model the route, not just the unit.
A comparison that holds up in a review
The table sets out the questions to ask of each model. Use it as a template, then fill in the numbers for your own product, volume and route.
| Question | Sub-assembly | Full assembly |
|---|---|---|
| Who owns the interface | You | The supplier |
| Piece price | Lower, excludes final build | Higher, includes build and test |
| Freight volume | Dense, more protection | Bulky, less handling |
| Inventory | Buffer of common parts | Buffer of finished units |
| Change speed | Fast at the final step | Every change is a supplier change |
| Quality visibility | You see incoming parts | You see the finished unit |
| Test ownership | Yours | Supplier, with records |
| Best fit | Variants and core process | Mature, stable, high volume |
The table is a decision aid, not a rule. A product can use both models at different tiers.
The hybrid that often wins
Many programmes settle on a hybrid. The supplier builds the module that needs a clean room, a press or a specialist test, and your plant does the final configuration and the customer specific labelling. The boundary sits where the test equipment and the variant logic live.
A second hybrid is kitting. The supplier assembles the difficult joints but ships the kit with the variant parts loose, and your line chooses the variant at the point of order. This keeps the supplier's volume high and your finished goods low, at the cost of a more complex line-side presentation.
A third is a tiered build. The supplier makes a tested sub-assembly, you make the harness and the enclosure, and a second supplier does the final build. That can work, but it needs a clear drawing at every boundary and a single owner for the tolerance stack. Without that owner, each tier blames the next.
What to put in the RFQ
State the boundary explicitly: which parts are in scope, which are free issue, and which are your responsibility. Include the assembly drawing, the bill of materials, the test specification and the packaging requirement. If the supplier is to own the interface, say so and ask for their stack-up and their proposed inspection points.
Ask for the cost broken down by material, purchased parts, labour, fixture amortisation, test, packaging and freight. A single unit price hides the assumptions you need to compare models. Ask what volume the price assumes, and what happens at half that volume.
Finally, ask how the supplier would handle a change. The answer tells you more about the working relationship than the price does. A supplier who can describe a change process, a first article and a cut-in date is ready to own a boundary. One who says it will be fine is not.
- Scope and boundary drawing
- Bill of materials with free issue items
- Test specification and acceptance limits
- Packaging and line-side presentation
- Volume, variants and release size
- Change process and first article expectation
Frequently asked questions
Is full assembly sourcing always cheaper?
No. It can be cheaper at high volume with a stable design, because the supplier spreads fixtures and learning over more units. It can be more expensive when variants are frequent, when freight is bulky or when the tariff classification changes. Compare total cost per good unit, not piece price.
Who owns the tolerance stack when a supplier builds a full assembly?
The supplier should own it, but you should review it. Ask for the stack-up, the datum scheme and the inspection points. Keep the interface drawing under your change control so a supplier change cannot move a critical dimension without your knowledge.
What is free issue and when should I use it?
Free issue means you supply a component to the assembler rather than the assembler buying it. It can protect a critical part or a price, but it adds logistics, duty and quality risk at the boundary. Use it when you have a strong reason, not as a default.
How do I choose between kitting and a full assembly?
Kitting suits frequent variants and low finished goods stock, because the variant is chosen late. A full assembly suits a stable product where the supplier can test the unit and ship it directly. The choice usually follows the number of variants and the value of the final test.
Sources and further reading
- International Chamber of Commerce, Incoterms 2020
- ISO 9001:2015, Quality management systems, requirements
- Qumek assembly sourcing models and programme reviews
Qumek Engineering Desk
Engineering team
The engineering team behind Qumek drawings, DFM notes, tolerances and inspection standards. Every guide is reviewed against the drawings and processes used on live programmes.
Test a sourcing boundary before you tender
Send the assembly drawing, the bill of materials and the volume. The engineering desk will map the interfaces and compare the two models on total cost.
Send a bill of materials