When a category manager sits across from a supplier and hears a price of $4.87 per unit, the natural question is: "Is that fair?" Most buyers answer that question by comparing it to last year's price, to what competitors charge, or to a target set by finance. None of those answers are a cost model.
Should-cost modeling answers the question differently. It builds the price from the ground up — raw material, direct labor, machine time, overhead, logistics, margin — and arrives at an independent number. The resulting gap between should-cost and quoted price becomes the basis for negotiation. Aranca documented a ~15% cost reduction on a specialty chemical after applying this methodology (Aranca, 2025).
The precise definition: what should-cost modeling is and is not
Should-cost modeling estimates the "fair" price of a product by decomposing it into its fundamental cost drivers and summing them bottom-up. It originated with the U.S. Department of Defense during Cold War procurement, when the Pentagon needed a way to challenge contractor pricing on complex weapons systems (Galorath).
Three related terms often get confused: does-cost is what you currently pay. Should-cost is what the part should cost given efficient production at current market rates. Could-cost is the theoretical minimum under perfect conditions — tighter tolerances, optimal plant location, best-in-class yields. Should-cost sits in the practical middle, not the theoretical floor. It is not a supplier margin attack. It is not a single number — it is a scenario range. And it is not a replacement for supplier relationships. It is a fact base for a better conversation.
Step 1: define scope and assemble the cross-functional team
Pick a part or assembly with high annual spend and at least moderate complexity — simple commodity items already trade at thin margins, so should-cost adds less value. Define the plant location, expected annual volume, and time horizon. Volume assumptions matter because they drive machine utilization and overhead absorption rates.
The fatal mistake at Step 1 is assigning this to procurement alone. Should-cost requires engineering (BOM, tolerances, process steps), production (cycle times, yields, scrap rates), and finance (labor rates, overhead allocation). Without at least these three functions, the model will miss cost drivers that a supplier sees clearly. Convergentis notes should-cost remains "underutilized and rarely trained" outside specialized cost-engineering teams, precisely because organizations skip the cross-functional step.
Step 2: build the bottom-up cost structure
Map the bill of materials down to the raw material grade and form factor. For each component, determine the material cost at current market rates — commodity indices for metals and resins, regional benchmarks for less standardized inputs. Map the manufacturing process sequence: operations, machines, cycle times, setups, changeovers, labor grades, and expected yields at each step.
Build each component separately before summing. This lets you isolate which cost drivers explain most of the gap — typically, material alone accounts for 50–70% of total cost in discrete manufacturing.
Step 3: calibrate with benchmarks and build scenario ranges
A model built in isolation is an academic exercise. Calibrate each input against internal historical purchase data, industry norms, and market indices. If your modeled labor rate is $28/hour but the regional average for that skill grade is $22, either you have a justification or the model is wrong.
This is where McKinsey's warning becomes critical: the final number matters less than the assumptions behind it (Fairmarkit). Build three scenarios: optimistic (best-in-class yields, favorable material pricing), baseline (realistic current conditions), and conservative (higher scrap, higher logistics, lower OEE). Present the range to internal stakeholders and to the supplier. A range is credible. A single number is brittle.
Step 4: negotiate from the fact base, not against the supplier
The most common failure in should-cost modeling — documented across multiple sources (Galorath, Fairmarkit) — is using the model as a blunt weapon. A buyer walks in with "our model says this should cost $4.12" and demands the supplier meet it. The supplier, who knows their actual costs, sees a model built on public data and assumptions that miss plant-specific realities.
The right approach: share the model's structure (not necessarily every input), ask the supplier where the assumptions diverge from reality, and use the gaps to identify joint cost-reduction opportunities. A die-cast part might be cheaper if redesigned for a different gate location. A machined component might benefit from a different raw material form factor that reduces cycle time. These structural savings — design-to-cost, process change, volume commitment — typically exceed what pure price negotiation achieves.
Step 5: integrate the model into ongoing category management
A should-cost model built for a single sourcing event and then shelved loses most of its value. Update material inputs when commodity indices move. Recalibrate labor and overhead annually. Track actual purchase prices against the should-cost benchmark and flag deviations. aPriori and similar platforms automate much of this, but the principle holds even with spreadsheet models: a maintained model is a category management asset. A one-off model is wasted effort.
Where should-cost modeling fails: the collapsed-step problem
The single most common failure mode is collapsing Steps 2 and 3 into a quick model built from desk research without engineering input. A procurement analyst Googles material prices, estimates cycle times from a video of a similar process, and calls the result a should-cost. This model has zero credibility with the supplier and will be dismissed in the first negotiation session.
A second failure: treating the model as evergreen. Material indices move. Labor rates change. A model built in Q1 2025 that has not been recalibrated by Q2 2026 is already wrong. Facton notes that static models become obsolete quickly in volatile commodity environments — exactly the environments where should-cost is most valuable.
A third: applying should-cost where transparency is structurally low. Service categories, IP-heavy components, and categories with highly customized processes resist bottom-up modeling. Fairmarkit advises that should-cost works best on categories where the physical transformation process is well-understood and data is available.
What correct execution looks like
Organizations that use should-cost effectively share five characteristics. They maintain dedicated cost-engineering roles that bridge procurement, engineering, and finance — not as a full-time job for every category, but as a capability they can deploy on high-value sourcing events. They update models on a fixed cadence, not ad hoc. They share model structure with suppliers and treat gaps as joint problems, not supplier failures. They integrate should-cost outputs into make-or-buy analysis and design-for-manufacturability reviews, not just price negotiations. And they report savings in terms finance recognizes — actual purchase price variance against a documented benchmark, not cost avoidance.
Operational checklist: before your next should-cost project
- Assemble engineering, production, and finance representation — not procurement alone
- Select a part with annual spend above $500K and well-understood manufacturing process
- Build the BOM to raw material grade level, not category level
- Source material prices from current market indices, not last year's purchase data
- Construct three scenarios: optimistic, baseline, conservative
- Share model structure (not necessarily every input) with the supplier before the negotiation session
- Document all assumptions — the assumptions are the model's real output
- Schedule quarterly recalibration of material inputs
What this means in practice
Audit your last three major direct material sourcing events. For each, ask whether you had an independent cost estimate before the first supplier meeting. If the answer is no — and for most procurement teams, it will be — pick the highest-spend part and build a should-cost model. Even a rough model changes the negotiation dynamic. You shift from "is this price competitive?" to "here is what we believe this should cost, and here is why."
Track the delta between should-cost and actual purchase price over the next 12 months. If the gap is under 5%, your category is well-priced and should-cost added analytical rigor. If the gap exceeds 10%, you have a sourcing priority.
Sources
- Aranca — Should Cost Model Case Study. Accessed June 26, 2026.
- Galorath — Should Cost Analysis Methodology. Accessed June 26, 2026.
- Fairmarkit — Should-Cost Analysis Glossary. Accessed June 26, 2026.
- Convergentis — The Power of Should-Cost Modelling in Procurement. Accessed June 26, 2026.
- aPriori — Should Cost Analysis Guide. Accessed June 26, 2026.
- Facton — Mastering Should Cost Analysis. Accessed June 26, 2026.
- Planergy — Should-Cost Modelling Guide. Accessed June 26, 2026.
- Precoro — Should-Cost Model: How to Build, Calculate, and Negotiate. Accessed June 26, 2026.