Should-Cost Analysis: How Buyers and Shops Build a Bottom-Up Cost

Every quote tells you what a supplier wants to charge. It does not tell you what the part actually costs to make. The difference between those two numbers is margin, overhead allocation, and — sometimes — padding that nobody questions because nobody rebuilt the cost from the ground up. Should-cost analysis closes that gap: it reconstructs the expected cost of a part from material, process, and time, so a buyer can challenge a quote with numbers instead of a feeling, and a shop can defend its price with a model instead of a shrug.

It is the same arithmetic whether you sit on the buying side or the selling side. A procurement engineer uses it to test whether €48 for a turned bushing is fair or fat. A shop owner uses it to sanity-check a quote before it goes out, and to answer the customer who emails back "your competitor is 20% cheaper." This article shows how to build a should-cost from the bottom up, where the numbers come from, and how to read the gap when the model and the quote disagree.

What should-cost analysis actually is

Should-cost — sometimes called "clean-sheet costing" or "cost teardown" — is a bottom-up estimate of what a part ought to cost a reasonably efficient supplier, built from first principles rather than from a price list. You are not asking "what did they quote?" You are asking "given this drawing, this material, this process, and a normal shop running at normal rates, what does each step add up to?"

The output is a cost breakdown, not a single number. That structure is the whole point. A blended price of €48 is impossible to argue with. A breakdown that says €11 material, €19 machine time, €9 setup amortized, €5 overhead, €4 margin is something both sides can examine line by line. The disagreement, when it comes, lands on a specific row — and that is a conversation, not a standoff.

The bottom-up cost stack

A should-cost model uses the same cost blocks any shop uses to build a price. If you have read how to price CNC machined parts, this stack will look familiar — should-cost is that same build, done by someone reconstructing the price rather than setting it.

Cost block	What it captures	Main driver
Material	Stock purchased, including waste and certification	Stock shape, grade, removal volume
Machine time	Cycle time × the fully burdened rate for that machine	Geometry, tolerance, machinability
Setup, amortized	Setup time × rate, divided by batch quantity	Batch size, fixture complexity
Overhead	Inspection, packaging, admin not already in the machine rate	Shop structure
Margin	Profit the supplier is entitled to earn	Risk, volume, relationship

The total is the first four blocks added together; the price is that total with margin applied. Keeping margin as its own visible row matters, because the most common mistake in reading a quote is treating a high price as "greedy margin" when the real cause is an underestimated cycle time or an expensive certified material. Separate the rows and you can see which one is actually large.

Building a should-cost model, step by step

The method is a checklist, not a flash of insight. Work it in order and the number falls out.

Read the drawing for material. Find the grade and stock shape in the title block. Calculate the stock volume the part consumes — finished volume plus the waste from facing, cutoff, clamping, and roughing. Multiply by price per kg. Add any certification premium (3.1 cert, traceability) the drawing demands. Do not use finished-part weight as a shortcut unless the part is nearly solid.
List the operations. Walk the geometry and write down every operation the part needs: turning, milling, drilling, tapping, second setups, deburring, inspection. Each one carries time.
Estimate cycle time per operation. Use realistic feeds and speeds for the material. Hard materials — tool steel, titanium, Inconel — run at a fraction of aluminum's surface speed, so the same geometry takes longer and costs more.
Apply a fully burdened machine rate. Use a rate built for that class of machine, not a shop average. A five-axis center is not a 2005 two-axis lathe. The machine shop hourly rate calculation method gives you defensible rates from depreciation, floor space, power, and labor burden.
Amortize setup over the batch. Setup is fixed per job. Estimate it in minutes, multiply by the rate, divide by quantity. At a batch of one it can double the unit cost; at fifty it nearly vanishes.
Add overhead, then margin. Layer in inspection, packaging, and admin if they are not already in the machine rate. Then apply a normal margin — 15% to 35% is a typical band for job-shop work — using the margin, not markup, formula: price = cost ÷ (1 − margin %).

A worked example

Take a small turned steel bushing, batch of 25, machined on a CNC lathe billed at €60/hour, with a five-minute cycle and a 50-minute setup.

Line	Calculation	Amount
Material	0.35 kg stock × €4.20/kg	€1.47
Machine time	5 min ÷ 60 × €60	€5.00
Setup, amortized	(50 min ÷ 60 × €60) ÷ 25 parts	€2.00
Overhead	inspection + packaging, ~12% of subtotal	€1.02
Total cost		€9.49
Margin (25%)	€9.49 ÷ (1 − 0.25)	+€3.16
Should-cost price		€12.65

If the supplier quotes €13, you are within noise — the price is fair and there is nothing to negotiate. If they quote €22, the model tells you exactly where to ask questions: is the cycle time really five minutes, or is there a second op you missed? Is the material certified to something pricey? Are they amortizing setup over a batch of five because they expect to re-set the job? You are no longer haggling over a total — you are testing assumptions, one row at a time.

Reading the gap between should-cost and the quote

A divergence between your model and the quote is information, not an accusation. Three honest explanations usually cover it.

Your assumptions are wrong. The most common case. You assumed a single setup; the part needs two. You assumed mild steel; the drawing calls a stainless grade that work-hardens. You guessed batch quantity. Should-cost is only as good as its inputs, and a real shop knows its own process better than a clean-sheet model does. When the supplier explains the gap and the explanation is a real operation you missed, update the model — that is the system working.

Their cost is genuinely higher. A shop with old machines, a long setup process, or expensive certified material may legitimately cost more to produce the part. That is a supplier-fit question, not a cheating question — and it is exactly the kind of structural cost a teardown surfaces.

There is padding. Sometimes the gap is just margin you can negotiate, especially on volume or a repeat relationship. The model gives you the standing to ask, because you can point at the row.

This is also why should-cost is useful inside a shop, not only against one. Running your own quote through the same teardown before it ships catches the estimate that drifted high or low, which is one of the most reliable ways of reducing quoting errors before they reach the customer.

The honest limits

Should-cost is a model, and models simplify. It assumes a "reasonably efficient" supplier, which no real shop is in every dimension. It struggles with proprietary processes, unusual tooling already amortized on another job, or a supplier whose volume on a material gets them pricing you cannot match. It is weakest on the things that do not show on a drawing: a shop's overhead structure, its current capacity, how badly it wants the work this month.

So treat the number as a reference, not a verdict. It is most powerful as a starting point for a conversation — a structured way to ask "help me understand this price" — and it is genuinely useful for distinguishing a quote that is honest and high from one that is simply padded. It does not replace knowing your suppliers. It also is not the same thing as a quote, an estimate, or a bid: should-cost is your private analysis, not a commitment to anyone.

A second teardown: a fabricated part, where the stack looks different

The bushing above is a machining teardown, and machining teardowns are the easy case — one machine, one rate, one cycle time. Most disputed quotes are for fabricated parts, where the cost stack tilts away from machine time and towards material yield and secondary operations. The method does not change; the rows that dominate do. Here is the same teardown run on a laser-cut, formed and coated bracket, batch of fifty.

Every figure below is an assumption chosen to make the arithmetic visible, not a benchmark — replace each one with your own before you take a number to a supplier.

Sheet price: assume €95 for one 2500 × 1250 × 3 mm mild steel sheet (an assumption, not a market price).
Nesting yield: assume 18 good parts per sheet (an assumption — this is the number to test first).
Laser time: assume 2.5 minutes per part at an assumed burdened rate of €100/hour for the machine.
Forming: assume 0.6 minutes per part on the press brake at an assumed €55/hour.
Setup: assume 45 minutes in total for the nest and the brake tooling, at the same assumed €100/hour, spread over the batch of fifty.
Coating: assume €1.80 per part bought in from a finisher (an assumption, including their handling).
Overhead and margin: the same treatment the bushing example above uses — overhead on the same basis as that example (there, roughly 12% of the subtotal) and margin at the same 25%, applied with price = cost ÷ (1 − margin). Both are carried over from this page, not introduced here.

Line	Calculation	Amount
Material	€95 ÷ 18 parts per sheet	€5.28
Laser time	2.5 min ÷ 60 × €100	€4.17
Forming	0.6 min ÷ 60 × €55	€0.55
Setup, amortised	(45 min ÷ 60 × €100) ÷ 50 parts	€1.50
Coating	bought in, per part	€1.80
Subtotal		€13.30
Overhead	same treatment as the example above	€1.60
Total cost		€14.90
Margin	€14.90 ÷ (1 − 0.25), the same 25% as above	+€4.97
Should-cost price		€19.87

Two things fall out of that table that never fall out of a machining teardown. First, material and cutting together are just under two thirds of the cost, so an argument about the hourly rate is an argument about the smaller half of the number. Second, the assumption carrying the most risk is not a rate at all — it is 18 parts per sheet. Move it to twelve and material per part goes from €5.28 to €7.92, and overhead and margin scale on top of it. When a fabricated quote comes back higher than your model, ask about the nest before you ask about the rate.

That is also the honest limit of a clean-sheet fabrication model: you are guessing at someone else's nesting, and they are not. If you want to test the yield and cut-length assumptions without rebuilding the table by hand, the laser cutting cost calculator takes cut length, pierces and sheet yield directly and shows which line moves. For the machining half of a mixed assembly, the cost blocks in how much does CNC machining cost map onto the same rows, and the rate you feed into either model should be built the way the machine shop hourly rate calculation sets out — burdened, per machine class, not a shop-wide average.

Run the teardown on both halves and the conversation with the supplier stops being about the total. It becomes a row-by-row comparison of assumptions, which is the only version of that conversation either side can win honestly.

From a drawing to a should-cost in minutes

The slow part of should-cost analysis has never been the arithmetic — it is the reading. Extracting material, operations, tolerances, and stock shape from a technical drawing by hand is the bottleneck, and it is exactly the step that gets shortcut under deadline, which is what makes a clean-sheet model wrong.

QuoteBuddy reads the technical drawing and surfaces the inputs a should-cost model needs — material callout, key dimensions, tolerance classes, and the operations implied by the geometry — then builds the cost up block by block from your machine rates, setup times, and target margin. Whether you are a buyer pressure-testing a supplier's number or a shop defending your own, you get a structured breakdown instead of a single total you have to take on faith.

Start a 30-day trial and run a few real drawings through it. Compare the bottom-up number it builds to the quotes on your desk — and see exactly which row the disagreement lives in.