Only 28.1% of procurement professionals know what should-cost analysis is. Only 4.6% actually use it. The gap between awareness and adoption is one of the widest in procurement, according to a CADDi survey cited by Art of Procurement. For the 95.4% who skip it, every supplier negotiation starts from the same weak position: the last price paid.
That is not a negotiation. That is a conversation about how much more the buyer will pay this year. Should-cost analysis replaces that conversation with an independent cost model that says: "Here is what this product should cost to make, based on materials, labor, overhead, and a reasonable margin. Now explain the gap between this number and your quote."
What should-cost analysis actually is
Should-cost analysis is a cost modeling method that estimates the total cost to produce a product or deliver a service, independent of any supplier quote. It breaks down every input — raw materials, direct labor, manufacturing overhead, tooling, logistics, and supplier margin — and assigns a defensible cost to each.
The output is not a target price. It is a baseline. The supplier's quote gets compared to the baseline, and the difference becomes the substance of the negotiation. If the supplier's quote is higher, the buyer asks: "Which cost element explains this gap?" If the supplier cannot answer, the buyer has leverage.
Step 1: Gather cost driver data, not just price history
Start by mapping every input that drives cost in the product or service. For a machined metal component, that means: raw material type and grade, weight, scrap rate, machine time per unit, labor hours, setup time, tooling amortization, surface treatment, packaging, and freight. For a service category, the drivers are labor rates, utilization assumptions, overhead allocation, and margin structure.
Cost drivers are not the same as cost elements. A cost element is "raw materials." A cost driver is "3.2 kg of 6061 aluminum at $4.15/kg with 12% scrap." One is a line item. The other is a model. Strategy & Business, in a survey of 17 core purchasing skills, found that capturing cost drivers — not just listing cost elements — was the single factor that separated effective cost models from generic ones.
Step 2: Build the independent cost baseline
Build the model with internal data and third-party indices first. Use published commodity prices for raw materials, Bureau of Labor Statistics data for labor rates by region, and industry benchmarks for machine rates and overhead multipliers. Do not ask the supplier for input yet. Step 2 is about building an independent reference point that no single supplier controls.
McKinsey's research on should-cost analysis found that most organizations see immediate savings from the method, but only a minority sustain them. The reason: they do not treat the baseline as a living document. Commodity prices shift. Tariffs change. Labor rates adjust. A model built in January is obsolete by March if it is not refreshed. The baseline is only as good as the data that feeds it.
Step 3: Engage supplier input — after the baseline exists
Share the baseline with the supplier. Ask: "Where does our model differ from your actual cost structure?" This is not an interrogation. It is a collaboration. Strategy & Business documented a McDonald's case study where a jointly developed chicken cost model — capturing mortality rates, weight gains, and feed mix optimization — produced results neither side could have achieved alone.
The supplier brings process knowledge the buyer's model cannot replicate: proprietary manufacturing techniques, volume-based raw material discounts, optimized machine scheduling. A collaborative model "has a greater probability of being fully applied," according to Strategy & Business. But the collaboration only works if the buyer arrives with a model. Without it, the conversation is: "Tell me your costs." With it, the conversation is: "Here is what we think. Correct us."
Step 4: Reconcile the gaps and negotiate from evidence
Every gap between the baseline and the supplier's quote is now a specific question. If the supplier's raw material cost is 8% higher than the commodity index, ask why. If their labor hours are 15% above the benchmark, ask about the process step that drives the difference. A gap that cannot be explained is a negotiation point. A gap that can be explained — a proprietary process that genuinely adds value — is not.
Inverto's methodology emphasizes that should-costing should not be used as "a means of exerting pressure." The goal is not to squeeze margins to zero. It is to identify where cost is justified and where it is not. Suppliers who understand their own cost structures respond to this. Suppliers who do not are revealed.
Step 5: Maintain and refresh the model over time
A cost model that lives in a spreadsheet on one analyst's laptop dies when that analyst leaves. McKinsey recommends establishing a permanent cost engineering function — cross-functional, tied to both procurement and product development — to maintain models as living assets. Honda reduced its cost-research function from 20–25 experts to approximately six by codifying cost expertise into reusable "cost tables" that outlived individual analysts.
For most organizations, the practical starting point is simpler: update raw material indices quarterly, review labor assumptions annually, and rebuild full models when the category goes back to market. Suplari found that models covering only the top 10–20 categories is the norm. The remaining 80% of spend goes to negotiation with nothing but last year's price. Extending should-cost principles to mid-tier categories — even with simpler, AI-assisted baselines — is where the next wave of savings sits.
Where most teams get it wrong: collapsing steps 2 and 3
The single most common failure mode is combining the independent baseline with supplier input into one step. A buyer asks the supplier for a cost breakdown, compares it to a rough internal estimate, and calls it should-cost analysis. This is not should-cost. It is supplier cost validation with a different name.
Without an independently built model first, there is nothing to validate the supplier's claims against. Every number the supplier provides becomes the baseline by default. The buyer has no leverage because they arrived at the table empty-handed. Suplari documented that traditional should-cost models cover only the top 10–20 categories and "go stale the moment commodity prices shift." When models are both narrow in coverage and stale in data, the method collapses entirely.
What correct execution looks like
Organizations that do should-cost analysis correctly share three characteristics. First, they have a dedicated cost engineering function — even if it starts as one person — rather than treating cost modeling as a project that gets handed to whoever is available. Second, they segment categories by modeling depth: full engineering models for the top 5–10 categories by spend, lighter AI-assisted baselines for the next 50–100, and anomaly detection for the long tail.
Third, they treat the model as a collaborative tool, not a weapon. Galorath's 2025 industry survey found that 63% of organizations have not implemented AI-driven estimation tools, despite 70% wanting real-time data and 71% wanting automation. The organizations closing this gap are not the ones throwing technology at the problem. They are the ones integrating cost models into ongoing supplier management, not just pre-negotiation preparation.
Operational checklist: should-cost analysis readiness
- Identify the top 5 categories by spend where a should-cost model would change the negotiation dynamic
- Map every cost driver for category #1: raw materials (grade, weight, scrap rate), labor (hours, rate), machine time, overhead, logistics
- Source independent commodity price indices and labor rate benchmarks — do not rely on supplier-provided data for the baseline
- Build the model internally before any supplier conversation. If a supplier sees it first, the baseline is theirs, not yours
- Share the model with the supplier as a collaboration tool, not an ultimatum. Ask: "Where does our model differ from your reality?"
- Reconcile every gap that exceeds 5% with a specific, documented explanation
- Schedule quarterly raw material index updates and annual full-model reviews
- Document the model methodology so it survives the analyst who built it
What this means in practice
Start with one category. Pick a direct material where raw material and manufacturing process costs are the dominant drivers. Build the model. Test it against historical pricing. Then bring it to the next supplier negotiation. The first model will take weeks. The second will take days. By the fifth, the methodology becomes repeatable.
For categories where a full engineering model is impractical — indirect spend, services, low-volume items — build a simpler cost driver map. Even identifying the three largest cost drivers and sourcing independent benchmarks for them changes the negotiation. The delta between "last year's price plus 3%" and "the raw material index is flat and labor rates are up 2%, so the increase should be 1.5%" is thousands of dollars per contract. Across a portfolio of 50 categories, it is margin that most procurement teams leave on the table.
Deloitte's 2025 Global CPO Survey identified capability gaps (40%) and talent gaps (34%) as the top barriers to delivering procurement value. Should-cost analysis addresses both: it builds institutional capability that does not walk out the door when an analyst leaves, and it develops the analytical muscle that separates procurement teams who influence P&L from those who process purchase orders.
How long does a should-cost model take to build?
A full engineering model for a complex direct material category takes 2 weeks to 2 months manually, according to Suplari. Simpler categories with fewer cost drivers can be modeled in days. The first model always takes the longest — subsequent models reuse methodology and data sources.
Do suppliers push back on should-cost analysis?
Suppliers push back when the model is used as a blunt instrument to demand price cuts. They engage when it is presented as a collaboration tool. Art of Procurement notes: "Even if you have a clear breakdown, if you do not have a good relationship with the suppliers, it doesn't make any sense." The relationship quality determines whether the model opens a conversation or shuts one down.
Can should-cost analysis be done without engineering expertise?
Full engineering models for complex manufactured components require engineering input. But lighter should-cost baselines — using commodity indices, labor rate benchmarks, and industry overhead assumptions — can be built by analytically strong procurement professionals. The key is knowing which cost drivers matter and sourcing defensible benchmarks for each.
Data sources
- Strategy & Business — "Cost Modeling: A Foundation Purchasing Skill". Accessed July 10, 2026.
- Art of Procurement — "Getting to Know Should-Cost Analysis" (citing CADDi survey data). Accessed July 10, 2026.
- Suplari — "Should-cost Modeling in Procurement". Accessed July 10, 2026.
- McKinsey & Company — "Find cost opportunities with today's should-cost analysis". Accessed July 10, 2026.
- Inverto (BCG) — "Fair Procurement Prices through Should Costing". Accessed July 10, 2026.
- Galorath — 2025 Industry Survey on AI-driven estimation tools. Accessed July 10, 2026.
- Deloitte — 2025 Global CPO Survey. Accessed July 10, 2026.