Should-cost modeling: why price-based negotiation fails for engineered components

Every procurement professional who buys engineered components knows the feeling: a supplier quote arrives, and the only reference point is last year's price plus whatever market intelligence the team assembled in a few days. The negotiation becomes a tug-of-war over percentage points — "we need 5%," "best I can do is 3%" — with neither side anchored to what the component actually costs to manufacture.

Price-based negotiation assumes the supplier's quote reflects cost reality. For commodity-grade materials with transparent market pricing, that assumption is reasonable. For engineered components — custom machined parts, injection-molded plastics, die castings, stamped metal assemblies, PCB sub-assemblies — it is structurally wrong. Suppliers have cost information asymmetry, and they use it. Research from aPriori and the DFMA Institute shows that organizations using should-cost analysis identify pricing gaps of 10-20% on modeled components compared to supplier quotes. The gap is not negotiation leverage. It is a data deficit.

Why price benchmarks fail for engineered parts

Price benchmarks compare what other buyers paid for similar components. That sounds useful, but for engineered parts the "similar" qualifier does a lot of work. A machined aluminum bracket with four drilled holes and a machined bracket with two tapped holes and a surface finish specification are fundamentally different products in cost terms, even if they look similar in a line-item description.

Cost drivers for engineered components are determined by specific manufacturing decisions: material grade and form (bar stock vs. plate, 6061 vs. 7075 aluminum), cycle time (which depends on part geometry, tolerances, and machine capability), tooling amortization (specialized fixtures vs. standard workholding), yield rates (scrap from setup and changeover), and finishing operations (anodizing, passivation, heat treatment, plating). None of these are captured in a price benchmark from a purchasing database.

A Galorath analysis of should-cost methodology notes that effective should-cost programs are cross-functional, involving procurement, engineering, manufacturing, and finance. This structure exists because the cost model requires manufacturing process knowledge — machine cycle time estimation, material utilization rates, labor efficiency assumptions — that procurement alone cannot provide. Price benchmarking alone cannot provide it either.

---

The information asymmetry that price negotiation cannot overcome

Suppliers of engineered components operate with complete visibility into their own cost structure. They know the machine they run the part on, the setup time, the cycle time, the tool wear rate, the material utilization percentage, and the overhead absorption rate. The buyer sees none of this. In a price-only negotiation, the supplier holds all the information and the buyer holds last year's invoice and a market index.

CostItRight's research on should-cost adoption in manufacturing notes that instead of asking for a general discount, buyers with should-cost models can discuss specific cost drivers — material cost, machining time, production processes — which shifts the conversation from "give me a better price" to "the material cost index has dropped 8% and your quoted price reflects pre-drop material. Let's talk about the delta."

This is the difference between negotiating margin and negotiating cost inputs. The first gets capped at whatever the supplier is willing to concede. The second targets specific, verifiable cost elements. They produce different outcomes.

---

What a bottom-up should-cost model actually looks like

A should-cost model for an engineered component starts with the bill of materials and works up from there. For a machined part, the model estimates the raw material cost by calculating the part volume, adding typical scrap allowances for the machining process (5-15% depending on complexity), and applying the relevant material price per unit weight. Then it calculates machining cost by estimating cycle time based on part features, machine hourly rates, and setup costs. Then it adds secondary operations: deburring, surface treatment, inspection. Overhead is allocated as a percentage of direct costs, typically 20-40% depending on the supplier's facility type.

DFMA's should-cost methodology emphasizes that the most effective programs start with 10–20 high-spend parts, build baseline models, and use them to structure supplier conversations. A cross-functional review cadence makes it repeatable. Organizations that attempt to model every component at once typically stall. The proven approach is selective coverage: model the 20% of parts that represent 80% of spend, then expand.

The Precoro should-cost modeling guide breaks the process into a clear pipeline:

---

When should-cost modeling works — and when it doesn't

Should-cost modeling performs best on components with high manufacturing costs relative to total cost — machined parts, castings, forgings, injection-molded plastics, stampings, and assemblies with multiple process steps. These categories have enough manufacturing content that the cost model reveals structural pricing differences, not noise.

The methodology performs poorly on categories where cost is dominated by proprietary IP, brand premium, or distribution markups that have no manufacturing analog. Software licensing, proprietary electronics modules with embedded firmware, and branded consumables fall into this category. Modeling manufacturing cost for a firmware-defined product tells you nothing about the value of the IP inside it.

GEP's analysis of should-cost modeling highlights that the accuracy of a should-cost model depends on the quality of its inputs. Organizations that invest in dedicated cost engineering teams — staffed with former manufacturing engineers and procurement cost specialists — consistently produce more accurate models than teams that rely on part-time analysts using spreadsheets. The gap between a well-sourced model and a spreadsheet estimate is typically 5-10 points of accuracy.

---

What this means in practice

Building a should-cost capability does not require an enterprise software investment in week one. The most effective path is targeted and iterative:

• Select 10–20 high-spend engineered components. Start with the parts where you spend the most and have the least cost visibility. Model these with a cross-functional team. Do not attempt to cover every category.
• Build the model collaboratively. Involve engineering for CAD and process data, manufacturing for cycle time and yield assumptions, and finance for overhead allocation. A model built by procurement alone will miss process-level cost drivers. A model built by engineering alone will miss commercial benchmarks.
• Anchor negotiations on specific cost elements. Walk into supplier meetings with decomposed cost data. Say "the material index for 6061 aluminum dropped 6% this quarter, and your price is flat. Let's discuss material cost adjustment." This changes the conversation from adversarial discounts to data-backed collaboration.
• Refresh models quarterly. Material indices, labor rates, and exchange rates shift continuously. A should-cost model built once and filed away is a snapshot. A model refreshed quarterly is a negotiation tool. AI-driven cost intelligence platforms can automate this refresh cycle.
• Measure the gap, not just the savings. Track the delta between your should-cost estimate and the final negotiated price. If the gap widens over time, your model assumptions may be stale. If it narrows, your negotiation discipline is improving. This is the only metric that tells you whether your cost intelligence is working.

---

Frequently asked questions