A production line’s output and its accepted deliveries can tell different stories. Compare yield and unit cost using the same product, period and treatment of rework.
A factory can report rising throughput while its cost per accepted product also increases. More material may be entering the process, but a larger share may require rework, remain unfinished or fail acceptance. The production claim and the cost claim can both be accurate while describing different populations.
For a defence company evaluating a supplier’s scale-up, the useful comparison begins with what can be delivered and accepted. A headline capacity figure provides context. A decision about price or schedule needs a clearer account of the product, the period, the acceptance point and the resources used to reach it.
Throughput can refer to work passing a particular point in a process. Accepted output refers to items meeting the applicable release conditions. Those quantities may coincide in a simple process, but a buyer should not assume that they do. Some output may still require inspection, documentation or another production step before it can support delivery.
The KPI discussion in NIST IR 8188 separates throughput, yield, utilisation and unit cost. The paper concerns a continuous-process research simulation; its value here is the distinction between measures, rather than a prescribed accounting method for a defence production line.
Ask the supplier where its reported output is counted. The answer should identify the product and the process boundary. If the figure concerns an intermediate operation, the remaining work belongs in the delivery assessment. A buyer can recognise a real increase in that operation’s capacity without treating it as an equal increase in finished deliveries.
A yield percentage is incomplete without its numerator and denominator. It might describe items passing a step the first time, items eventually accepted after rework or a material conversion measure. These can be useful for different decisions. Comparing them as if they were identical can make one process appear better simply because it uses a different definition.
Consider an illustrative batch of 100 equipment modules. Eighty pass the stated acceptance process on the first attempt. Fifteen more are accepted after rework, and five remain rejected. First-pass acceptance is 80 per cent, while eventual accepted output is 95 modules. Reporting only the latter conceals the additional effort needed to recover the fifteen items.
The example does not prescribe a universal yield formula. It shows why the report should explain what happened to the same population. If some modules are still awaiting a decision, label them as unresolved rather than counting them as either successful deliveries or final losses prematurely.
Recovered output can be commercially valuable. The question is what it consumes and whether the plan includes those resources. Rework may require additional labour, material, equipment time or inspection. It can also occupy capacity that was expected to process new orders.
For the illustrative batch, suppose the initial production effort costs 10,000 currency units and the recovery work adds 1,500. If 95 modules are accepted, those stated costs average about 121.05 per accepted module. Dividing the initial 10,000 by the original 100 would instead produce 100, a different figure based on a different cost and output boundary.
Neither calculation describes a real company or a complete accounting treatment. The point is to make the assumptions visible. A quoted unit cost should identify whether it includes the work necessary to obtain accepted output, and how costs associated with rejected or unfinished items are treated.
A monthly input figure should not be divided into an output figure drawn from a different production cycle without explanation. Items started near the end of a period may finish later. Earlier work in progress may support this month’s deliveries. Timing can therefore change the apparent relationship between spending and accepted output.
Product mix also matters. A line producing several variants may experience a change in average cost because the mix changes, even if each variant’s process performs consistently. Ask whether the comparison concerns the same configuration and volume profile, or whether it has been adjusted to account for a different mix.
These questions are particularly relevant when a supplier moves from a development batch to repeat production. Early spending can include preparation that benefits later units. The analysis should explain which costs are recurring and which belong to the establishment of the route, rather than assuming every observed improvement will continue indefinitely.
GAO’s Cost Estimating and Assessment Guide overview emphasises the estimate’s purpose, technical baseline, assumptions, supporting data and updates using actual costs. It also identifies sensitivity and risk analysis. Those are useful disciplines when a company turns early production evidence into a commercial forecast.
For a purchasing comparison, establish what the proposed price includes: the product configuration, quantity, acceptance records, delivery conditions and any separately funded development work. A lower recurring price may be accompanied by a larger initial commitment. Another proposal may include more support or a different allocation of change risk.
The estimate should also identify the volume assumption. A price based on a sustained production run may not apply to intermittent orders. The customer can then decide whether its demand supports the proposed economics or whether it needs a quotation for a more realistic ordering pattern.
Rather than asking only for a single future cost number, identify which assumption most strongly changes it. The answer could be accepted yield, labour time, material price or the utilisation of a constrained resource. The supplier should explain which assumptions are supported by observed production and which remain targets.
A simple commercial comparison can examine a few stated cases using the same cost boundary. If expected accepted output falls, what happens to the cost per delivery and the promised schedule? If rework consumes more inspection time, is that resource available? These are planning questions, not predictions that a specific supplier will fail.
Our guide to manufacturing capacity evidence addresses the related constraint problem. A strong yield result at one step does not establish that all later activities can handle the same volume. The cost forecast and the delivery forecast should therefore use a compatible view of the whole route.
Even a well-supported cost estimate does not determine the selling price on its own. Commercial terms may allocate inventory exposure, warranty obligations or demand uncertainty differently. A buyer should use production evidence to understand the quotation, while comparing the complete commitment it receives. Requiring a supplier to disclose every internal cost is not always necessary; a transparent explanation of the assumptions governing price and delivery may answer the relevant purchasing question.
The buyer can ask for evidence appropriate to the decision it is making. A development order may establish the process and produce an initial cost baseline. A repeat order can provide evidence about variation and rework. A larger commitment can then be assessed against the accepted output actually achieved and the remaining dependencies.
The first article inspection guide explains another boundary: evidence about an initial configuration does not by itself establish sustained production economics. Keeping those questions separate allows useful progress to be recognised without overstating what the evidence covers.
A credible scale-up claim connects the quantity started, the quantity accepted and the cost of reaching that result. It explains unresolved items and the assumptions behind the next stage. That gives customers a firmer basis for agreeing price, schedule and volume than a single throughput figure detached from the product they intend to buy.
GAO’s public cost-guide overview and the KPI table in NIST IR 8188 were read. The NIST paper concerns a continuous-process research simulation, not a discrete-production accounting standard. Numerical examples below are illustrative BDI calculations.
Suggest a correction