Should-cost analysis: building a part price up from the process

Start from the process route, because everything follows from it

Decide how the part would actually be made: the forming operation, the machine class and size it needs, the number of cavities or stations, the secondary operations, the inspection required. That route determines material form and yield, cycle time, manning and tooling. Guessing the route produces a model that is precise and wrong. Where your team lacks the process knowledge, walking the part with a manufacturing engineer or a supplier prepared to explain its approach is a better investment than refining the arithmetic around a route nobody has validated. A route validated by someone who has run the equipment is worth more than a model refined by someone who has not.

Material is where most models go astray

The purchased weight is not the finished weight. Sprues, runners, trim, kerf, machining allowance, nesting efficiency and setup loss all sit between them, and their magnitude varies enormously by process. Then there is the form: bar, plate, coil or granulate carries a different price from the underlying commodity, and the quantity a small supplier buys carries a different price again from an index. Recovered scrap value offsets part of the loss for metals and rarely for filled polymers. Getting these details roughly right matters more than any refinement elsewhere in the model.

Machine rates and the manning that goes with them

An hourly rate should reflect the specific equipment: its capital cost, expected life, utilisation, floor space, power draw and maintenance, plus the operator attention it genuinely requires. A press attended by one operator running several machines carries different labour than a manual cell. Regional wage differences matter but less than teams assume, because a low-wage plant often runs older, slower equipment and more manual handling, which consumes much of the advantage. Modelling both the rate and the cycle honestly explains more price variation between countries than the wage figure alone.

Using the model without turning it into a weapon

A should-cost figure presented as the price a supplier must accept produces defensiveness and, at best, a concession that reappears elsewhere. Presented as a model with visible assumptions and an invitation to correct them, it produces a technical conversation that frequently benefits both sides: the supplier explains why its cycle is longer, and either you learn something about the part or they discover an improvement. The purpose is to make the cost drivers discussable. A supplier that can explain its deviation from your model line by line is usually telling you the truth.

What the model is for beyond negotiation

The same structure answers questions a quote cannot. It shows which design features carry the cost, so engineering knows where to work. It tests whether a process change or a different machine size would pay. It gives a defensible view of what a price should do when material indices move, separating the volatile element from the conversion element. And it provides a sanity check on an internal build estimate before a make-or-buy comparison, where in-house numbers are often assembled on a different basis from supplier quotations. It is also the only defensible way to judge whether a proposed price increase reflects an input movement or an attempt to recover margin lost elsewhere.

Frequently asked questions

How accurate does a should-cost model need to be?

Accurate enough to identify which elements drive the cost and to show where a quote departs from a sensible process, which is a much lower bar than predicting a price. Effort should concentrate on material form and yield, cycle time and the machine class, since those dominate. Refining overhead allocation or margin assumptions to a fine degree adds little, because those are the elements you will discuss with the supplier rather than determine independently.

Where do you get machine rates and cycle times without a supplier's help?

From your own plant where the process exists internally, from manufacturing engineers who have run comparable equipment, from equipment builders who publish cycle characteristics for their machines, and from published industrial statistics for wage and energy inputs by country. Building a small internal library of rates by machine class and region, refreshed periodically, is far more useful than reconstructing the assumptions from scratch each time a part needs analysing.

Does should-cost work for a purchased sub-assembly?

Yes, but it has to be built level by level, modelling the significant components and then the assembly labour, test and the assembler's overhead and margin on top. The exercise often reveals that you are paying margin twice on components you already purchase directly elsewhere, which is a negotiable finding. For modules with many small components, model the few that dominate the content and treat the remainder as a block rather than attempting every line.