Assembly cost model: labour, fixtures, test and yield

Why a cost model beats a price comparison

Two assembly quotes can differ by 30 per cent and both be honest. One supplier may have a faster fixture, a lower scrap rate or a shorter test. Another may be pricing a manual build with a high labour content in a market where labour is cheap but turnover is high. Without a model, the buyer sees a number and guesses at the reason.

A cost model does not replace the quote. It gives you a way to test the quote, ask a better question and see which lever would change the price.

The five lines of an assembly cost

Start with direct material: the purchased parts, the free issue parts and the consumables such as screws, adhesive, labels and packaging. Add the cost of any process material, such as potting compound or conformal coating, that is consumed in the build.

Next is direct labour. Estimate the cycle time for each step in seconds, including load, assemble, fasten, inspect and unload. Multiply by a fully loaded rate that includes wages, benefits, supervision, facilities and idle time. The loaded rate is often two to three times the basic wage, and it is the number that should be used.

Third is equipment. Fixtures, presses, testers, ovens and automated stations have a purchase cost, a life and a utilisation. Divide the cost by the number of good units the equipment will make over its life, then add maintenance and calibration. If the equipment is customer owned, the amortisation may sit with you rather than the supplier, but the cost still exists.

Fourth is test and inspection. Include the time for the test, the cost of the test equipment, the consumables and the labour to disposition failures. A functional test that takes 20 seconds is not free at volume.

Fifth is yield and scrap. Every step has a first pass yield, and the combined yield is the product of the step yields. A line that runs at 96 per cent at each of five steps is at about 82 per cent overall. The material and labour in the scrapped units has to be carried by the good units.

Cycle time is the number to argue about

Labour cost is the line most often estimated by feel. Ask for the cycle time by step, then watch the station. The difference between the quoted time and the observed time is usually handling, not assembly.

Handling includes picking, orienting, walking, opening a bag and finding a tool. A well designed station removes most of it. A kit that presents parts in the order of use, a fixture that orients the part and a tool that is already in reach can remove several seconds per unit, which is substantial at volume.

Learning matters at launch. A new build is slower in the first weeks, and a quote that assumes full rate from day one will either be wrong or will hide a launch cost. Model a ramp: for example, 70 per cent of standard rate in week one, 85 per cent in week two and full rate from week four. Label it illustrative and replace it with the supplier's actual data when it exists.

Fixtures and tooling

A fixture is a fixed cost spread over volume. The same fixture that costs little per unit at a million units can dominate the price at five thousand. The model should show the amortisation separately so the buyer can see what happens if volume falls.

Ask how many units the fixture is expected to make, how often it is serviced and what the replacement cost is. A fixture with a worn locator will produce bad parts before it is replaced, so maintenance belongs in the model as a running cost, not a one-off.

Tooling for moulded or stamped components is usually quoted separately. Keep it separate in the model, but include the amortisation in the unit cost if the buyer is paying for it through the piece price.

Yield, rework and the cost of a bad unit

Scrap is not only material. A scrapped assembly at the end of the line carries all the labour, test time and purchased parts that went into it. A scrapped unit at step two carries much less. The model should apply yield at the step where the loss happens, not as a single rate at the end.

Rework is different from scrap because it adds labour without adding output. If a joint can be reworked, the cost is the rework time plus the risk that the reworked unit is weaker. If it cannot, the unit is scrap. Both should be in the model, because a process with high rework can look acceptable on scrap alone.

Ask for the first pass yield, not the final yield. A supplier can sort and rework to reach a good final number while the first pass yield stays low. The first pass number is the one that reflects the process.

The model in a table

The table is a template for a single assembly. Fill it with the supplier's data, then test the sensitivity of the total to the two or three inputs that matter most.

Volume, learning and the price curve

Assembly cost falls with volume for three reasons: fixtures amortise, labour learns and purchasing gets better prices. The first is arithmetic, the second is real but finite, and the third depends on the supplier's buying power with its own vendors.

Model the volume breaks explicitly. Ask what changes between 10,000 and 50,000 units: a second cavity, an automated station, a different test method or simply a lower labour rate. A price that falls without a process change is a commercial decision, and it may not hold.

The learning curve should be agreed, not assumed. A common approach is to hold the launch price for the first quantity, then review at an agreed volume with the actual cycle time and yield. That is fair to both sides and avoids a permanent price built on a temporary inefficiency.

Using the model in a negotiation

Use the model to ask questions, not to accuse. If the labour line is high, ask for the cycle time by step. If the yield line is optimistic, ask for the first pass data. If the fixture line is large, ask for the life and the service plan.

Keep the model updated. Labour rates, material prices, freight and volume all move, and a model built at the RFQ stage is out of date by the second year. Date every assumption and review it at the annual price review.

Finally, compare the model with the landed cost. An assembly that is cheap at the supplier's dock can be expensive at yours once freight, duty, handling and safety stock are added. The two models belong side by side, because the cheapest assembly is the one that arrives at the lowest total cost per good unit.