AI Supply Chain BriefShould-cost model example: a filled-in build-up and an AI prompt to make your own
Short answerA should-cost model estimates what a part ought to cost by building it up from material, processing, setup, scrap, overhead, profit and freight, then comparing that to the supplier's quote. In the fictional example below, a molded housing quoted at $0.98 builds up to about $0.83, a $0.15 gap to explore in negotiation. The gap is a question to ask, not proof.
Fictional exampleExample Appliance Co., its suppliers and every number in this example are made up for illustration. Don't paste real supplier quotes, NDA pricing, personal data or export-controlled data into an AI tool. Anonymize first and follow your company's AI policy.
What is a should-cost model, and what isn't it?
A should-cost model builds up what a part ought to cost from its elements: material, processing, setup, scrap, overhead, profit and freight. DFMA, a maker of design-for-manufacture software, writes it as "Should-cost = Material + Processing + Setup + Tooling + Scrap / Yield + Overhead + Supplier Profit". US federal acquisition rules describe should-cost reviews as "a specialized form of cost analysis" that does "not assume that a contractor's historical costs reflect efficient and economical operation." Those rules apply only to US government contracting, but the idea carries over to commercial buying: the model estimates what the part should cost, not what the supplier says it costs.
A should-cost model isn't proof that a supplier is overcharging. It's a structured way to find the questions worth asking.
The worked example: a molded housing
Example Appliance Co. buys a polypropylene housing at a quote of $0.98 a unit, 50,000 units a year. The buyer paid for the mold separately, so tooling is excluded. All rates below are assumptions for the example, not market benchmarks. Use your own or published rates.
| Cost element | Category | Quantity | Rate (assumption) | Cost per unit |
|---|---|---|---|---|
| PP resin | Material | 0.12 kg | $2.40/kg | $0.288 |
| Scrap allowance (5% of resin) | Material | 0.006 kg | $2.40/kg | $0.014 |
| Molding: 30 s cycle on 2 cavities = 15 s per part | Processing | 0.00417 hr | $60/hr (machine + operator) | $0.250 |
| Setup: 3 hr per 5,000-part run | Setup | 0.0006 hr | $60/hr | $0.036 |
| Packaging | Material | 1 | $0.03 | $0.030 |
| Manufacturing cost | $0.618 | |||
| SG&A overhead (15% of manufacturing) | Overhead | 15% | $0.093 | |
| Supplier profit (10% of cost + SG&A) | Profit | 10% | $0.071 | |
| Freight to plant | Logistics | $0.050 | ||
| Should-cost | ≈ $0.83 |
- Gap: $0.98 − $0.83 ≈ $0.15 a unit (about 15% of the quote), or about $7,400 a year at 50,000 units.
- Read the gap: the processing line has the biggest swing. If the supplier runs a slower cycle or a 1-cavity tool, the gap shrinks or disappears, so ask before you negotiate.
- Sensitivity: at a 40 s cycle on 2 cavities, molding becomes 20 s a part = $0.333. Manufacturing becomes $0.702, plus 15% SG&A = $0.807, plus 10% profit = $0.888, plus $0.05 freight = ≈ $0.94. One assumption closes most of the gap.
How do you turn the gap into negotiation questions?
Ask about the assumptions, not the margin. For example:
- "Our estimate of the cycle time is 30 seconds on 2 cavities. What's yours?"
- "What part weight and resin grade did you quote, and what resin price is that based on?"
- "What run size and setup time are in your price? Would a larger release quantity lower it?"
- "Is any tooling cost included in the piece price? We paid for the mold separately."
If the supplier's answers match your assumptions, the gap is real and worth discussing. If they don't, update your model. Either way you've learned something.
What are the common should-cost mistakes?
- Counting overhead twice. DFMA warns against adding overhead that the machine charge-out rate already includes. Know what's in your machine rate.
- Charging tooling you already paid for. Check whether the mold or fixture was bought separately.
- Wrong units. Cycle time per shot vs. per part, grams vs. kilograms, hours vs. seconds.
- Stacking profit. Apply profit once, not on every line.
Do it with AI
You are a cost engineer helping a buyer. Build a should-cost table for [PART] using only the inputs I give: [WEIGHT, MATERIAL PRICE, CYCLE TIME, CAVITIES, MACHINE RATE, SETUP, RUN SIZE, OVERHEAD %, PROFIT %, FREIGHT]. One row per cost element, with columns quantity × rate = cost per unit. Apply overhead and profit once. Where an input is missing, write 'NEEDED' and do not estimate it. Then compare to the quote of [PRICE], name the two assumptions that move the total most, and write 3 questions for the supplier.
What good output looks like: a table like the one above, with "NEEDED" wherever you didn't supply a rate, the two assumptions that move the total most (here, cycle time and cavities), and three specific questions.
What to check: recompute the table in Excel, make sure the machine rate isn't already loaded with the overhead you're adding, and confirm whether you've already paid for the tooling. Don't paste a supplier's confidential cost breakdown into an unapproved tool.
Template layout
Use these columns in any spreadsheet: cost element · category · quantity · rate · cost per unit · source or assumption. The last column matters most. Every rate should say where it came from, so anyone can challenge it.
A should-cost model is also your strongest backup when a supplier asks for more money. See how to respond to a supplier price increase. For more negotiation prompts, see ChatGPT prompts for procurement and the supplier negotiation and RFQ and sourcing events use cases.
FAQ
What does should-cost mean in procurement?
Should-cost is an estimate of what a part or service ought to cost, built up from its cost elements (material, processing, setup, scrap, overhead, profit, freight) instead of taken from the supplier's quote or past prices. Buyers use the gap between should-cost and the quote to plan negotiation questions.
Can you show a should-cost model example?
Yes. This page builds up a fictional polypropylene housing line by line: resin, scrap, molding time, setup, packaging, overhead, profit and freight. It comes to about $0.83 against a $0.98 quote, and a sensitivity check shows that a slower cycle time would close most of the gap.
Is there a should-cost model template for Excel?
A simple one works well: columns for cost element, category, quantity, rate, cost per unit, and source or assumption. Use one row per cost element, apply overhead and profit once, then add freight. The worked table on this page follows that layout.
How is should-cost different from price analysis or cost analysis?
Price analysis compares a price to other prices, such as competing quotes or past purchases. Cost analysis looks at the elements behind a price. US federal acquisition rules call should-cost a specialized form of cost analysis that doesn't assume the supplier's historical costs are efficient.
Do I need should-cost software?
Not for simple parts. A spreadsheet with your own or published rates is enough to find the big assumptions. Dedicated software helps with complex parts, many SKUs or processes where you lack in-house cost data.
Can AI build a should-cost model?
It can structure the model, apply your rates consistently, point out the assumptions that matter most and draft supplier questions. It shouldn't supply the rates, so tell it to write NEEDED for anything missing. Then check the math yourself.
