The hook: “Three bids and a buy” is lazy procurement
If three suppliers quote you $9.80, $10.10, and $10.25, you didn’t “validate the price.” You validated that they live in the same cost reality—or that they talk to the same customers, use similar quoting templates, and know roughly what the market will tolerate. Competitive bidding has a place, but using it as your only logic is a way to outsource thinking.
Should-cost analysis is the antidote: you build an independent estimate of what a part or service should cost to produce, then compare it to the quoted price. The point isn’t to accuse suppliers of overcharging. The point is to separate real cost drivers (material, cycle time, yield loss) from padding, risk premiums, and “because we can” margins.
The uncomfortable truth: should-cost doesn’t always produce savings. Sometimes it tells you the supplier is already efficient, or that your spec choices (tight tolerances, exotic finishes, short lead times) are the real problem. That’s still a win because it gives you options besides haggling.
The anatomy of cost: what you’re actually modeling
A usable should-cost model is not a PhD thesis. It’s a structured estimate that gets the big rocks right and makes assumptions explicit. For manufactured goods, a practical breakdown is: raw materials + conversion (labor and machine time) + overhead + scrap/yield + logistics/packaging + reasonable profit.
1) Raw materials: start with mass, grade, and yield
Material is where many models go wrong because buyers use the finished part weight and forget process yield. A machined part might start as bar stock with 30–70% removed. An injection-molded part might have runners and sprues. A stamped part might have nesting loss. Your estimate needs “buy weight,” not just “part weight.”
Define material spec (e.g., 6061 aluminum, 304 stainless, PA66 GF30). If you can’t name the grade, you can’t price it.
Estimate buy weight: finished weight ÷ (1 − scrap rate). Use a rough scrap assumption if you don’t know; write it down and refine later.
Apply a market reference price (index or published range) and add realistic conversion adders only when appropriate (e.g., alloy surcharge, plating chemistry, resin color masterbatch).
2) Direct labor and machine time: cycle time is the battleground
Suppliers rarely “make money” on hourly rates; they make money on time assumptions. A quote that bakes in a 6-minute cycle when the process should run at 2 minutes is a quiet margin multiplier.
Model conversion cost with two numbers: cycle time per unit and an all-in shop rate. If you don’t know the exact process, pick the likely one and sanity-check it. Example: CNC machining (minutes of spindle time), stamping (hits per minute + changeover amortized), molding (seconds per shot × cavities), assembly (touch time).
Direct labor cost = touch time (hours/unit) × loaded labor rate ($/hour).
Machine cost = machine time (hours/unit) × machine rate ($/hour).
Amortize setup/changeover when it’s meaningful: setup hours × rate ÷ batch size.
3) Overhead: don’t let it become a dumping ground
Overhead is real (maintenance, supervision, utilities, quality, depreciation), but it’s also where sloppy quoting hides. In should-cost, you don’t need their exact overhead pool; you need a defensible proxy. Many categories can be modeled using an overhead factor applied to direct labor and/or machine cost, or a fully burdened shop rate that already includes overhead.
Be careful with double counting: if you use a “shop rate” that includes overhead, don’t add an extra overhead percentage on top. The most common mistake is stacking burdens until the model becomes a weapon instead of a tool.
4) Scrap, yield, and rework: the hidden tax
Scrap isn’t only material loss; it’s wasted labor and machine time. If a process has a 95% first-pass yield, you effectively need to run 1.053 units of effort to ship 1 good unit. For complex parts, yield assumptions can swing cost more than raw material prices.
5) Packaging, logistics, and duties: small numbers that add up
Packaging and freight are easy to ignore until they become the negotiation excuse (“We can’t move price because freight”). Put them in your model explicitly: packaging per unit (bags, trays, pallets), inbound freight assumptions (mode, distance), and any known duties. You’re not trying to be perfect—you’re trying to prevent surprises.
6) Reasonable profit: yes, include it
A should-cost model that assumes zero profit is a fantasy and suppliers will treat it as hostile. Include a profit line. The right number depends on risk, capital intensity, and market dynamics. A stable, high-volume, well-specified part should carry a lower margin than a low-volume, high-uncertainty job with volatile inputs.
Your goal isn’t to dictate their P&L. It’s to arrive at a price that makes sense given the work content and the risk you’re asking them to carry.
The data hunt: proxy data you can get without an engineering degree
Mid-level buyers get stuck because they think should-cost requires proprietary factory data. It doesn’t. You can build a credible model with proxy inputs, then tighten it through supplier dialogue and internal learning.
Commodity and material references
Use publicly available commodity indices and published market references as anchors, then adjust for form and purchase size. For metals, you’ll often need to separate base metal pricing from processing premiums (e.g., mill forms, alloy surcharges). For plastics, resin pricing references and supplier quotes for standard grades are often enough to bracket the number.
Internal: your own PO history by material grade and form (bar, sheet, coil, resin).
Public: commodity exchanges and published index providers (use as directional anchors, not exact invoice prices).
Distributor quotes: quick budgetary pricing for common forms and quantities.
Supplier should-cost “build-ups” from prior negotiations—reuse carefully and update assumptions.
Labor and shop rate proxies
You rarely need to know a supplier’s exact wage table. You need a local labor cost range and a plausible burden. Government labor statistics, industry salary surveys, and regional manufacturing wage benchmarks can give you a starting point. Then translate that into a loaded rate (wages + benefits + payroll taxes) and compare it to the shop rates implied by quotes.
If you’re buying globally, treat labor rate as only one variable. Productivity, automation, scrap, and freight can erase “cheap labor” quickly—especially for bulky or low-value items.
Process time estimates without process mastery
You don’t need to be a manufacturing engineer to estimate time; you need to ask better questions and use simple heuristics. Start with: what process is most likely, what drives cycle time, and what would be an aggressive-but-plausible time versus a conservative one.
Use drawings and specs to identify complexity drivers: tight tolerances, surface finish, secondary ops, inspection requirements.
Ask internal SMEs (quality, manufacturing, R&D) for a 10-minute sanity check—one short meeting can save weeks of bad assumptions.
Request process assumptions from suppliers: cavity count, cycle time, yields, setup time, inspection plan. You’re not asking for trade secrets; you’re asking for the logic behind the number.
Build a range (best case / expected / worst case) instead of one “magic” cycle time.
A simple model template you can run in a spreadsheet
Keep the spreadsheet readable. If your future self can’t audit it in five minutes, it will be ignored during negotiations.
Inputs tab: material grade, buy weight, scrap/yield, cycle time, setup time, batch size, labor rate, machine rate, packaging, freight, profit %.
Calculation tab: material cost, conversion cost, scrap factor, overhead (if separate), logistics, total cost, profit, should-price.
Sensitivity tab: change resin/metal price, yield, cycle time, and see the price move. This is what you’ll use in the meeting.
At the negotiation table: forcing transparency without torching the relationship
Walking in with a should-cost model and announcing “You’re 18% too expensive” is a good way to get a polite smile and a defensive supplier. The better move is to use the model as a shared problem statement: “Here’s how we think the cost is built. Tell us what we’re missing.”
How to present the model
Lead with assumptions, not accusations: “We assumed 2.5 min cycle time and 92% yield based on similar parts.”
Show the biggest drivers first (usually material, cycle time, yield). Avoid arguing over pennies like box costs unless they’re truly material.
Use ranges: “If cycle time is 2–4 minutes, price should land between X and Y.” Ranges reduce ego battles.
Ask for their logic: “What cycle time are you quoting? How many cavities? What’s the inspection content?”
Where negotiations actually move
You’ll get more traction by changing the cost structure than by demanding a discount. Examples that routinely work: adjusting batch sizes to reduce setups, relaxing non-functional tolerances, switching to an equivalent material form that reduces scrap, agreeing to indexed material pricing with a transparent adder, or funding a tooling change that cuts cycle time.
A counterintuitive truth: sometimes paying more is the right call if it buys you yield stability, shorter lead times, or lower total cost of quality. Should-cost helps you see when a “cheap” quote is cheap because someone is ignoring risk.
Common supplier responses—and how to handle them
“Your material price is wrong.” Ask for their purchase form (coil vs sheet vs bar), minimum order quantities, and any surcharges. Update your model live if the evidence is credible.
“Cycle time can’t be that low.” Ask what constraint drives it (cooling time, tool wear, manual handling, inspection). Constraints are negotiable via process or spec changes.
“Overhead is higher here.” Separate legitimate overhead (regulated environment, energy costs, quality systems) from vague claims. Ask what the overhead covers that your model doesn’t.
“We can’t open our books.” You don’t need their books. You need the operational assumptions behind the quote. Keep the request focused on drivers, not their margin.
Protect the relationship while still being firm
Treat should-cost as a joint fact-finding exercise. Make it clear what you will do with the outcome: award volume, extend term, dual-source, redesign, or index pricing. Suppliers cooperate when they see a path to predictable business, not a one-time squeeze.
If the supplier refuses to discuss drivers and only repeats “that’s our price,” you’ve learned something useful: you’re buying in a market where either (a) switching costs are high, (b) capacity is tight, or (c) you don’t have enough alternatives. Should-cost won’t fix those realities, but it will stop you from pretending a fourth quote would.