Should-cost analysis: the complete guide (with examples)

Also called cleansheet or zero-based costing, should-cost is the discipline of computing what a product or service SHOULD cost to produce and sell profitably, without relying on the supplier's price as an anchor.

The output is a target price and — more importantly — visibility into the cost drivers that actually move the needle.

High-spend items (top 20% of spend by category), specification-driven products (packaging, machined parts), commoditized services with opaque rate cards (agencies, IT services), and any renegotiation after 24+ months without a market check.

It's overkill for small tail-spend and for services with published rate cards.

1. Raw materials: the resin, steel, corrugated, or labor category, at current market index price × yield loss.

2. Direct labor: minutes per unit × loaded labor rate for the region.

3. Manufacturing overhead: energy, tooling, depreciation, maintenance, allocated per unit.

4. SG&A + logistics: freight, warehousing, sales cost.

5. Margin: what a healthy supplier in this segment should earn (typically 8–18% depending on category).

A 40×30×25cm double-wall box: corrugated at 620 USD/ton, weight 480g/unit → ~$0.30 raw material. Add 4c die-cutting labor, 3c print, 5c overhead, 2c freight, 15% margin → target ~$0.51/unit.

If the incumbent quotes $0.68, you have a defensible 25% negotiation target — and you know it's split between labor efficiency and margin.

Use these as sanity checks when you sanity-test your target price. If your model implies a supplier margin far outside the band, revisit assumptions before showing it externally.

Corrugated & flexible packaging: 6–12%. Injection-molded plastics: 8–14%. Machined metal parts: 10–18%. Contract manufacturing (electronics): 5–10%. IT services & staffing: 10–20%. Marketing agencies: 15–25%. Freight & 3PL: 3–8%. MRO distribution: 20–30% (they carry inventory risk).

"Based on our own bottom-up model — using current [material index] at [X], regional loaded labor at [Y] and typical yield of [Z] — we're seeing a target price around [target]. We're not asking you to match our number blindly; we want to understand where our assumptions differ from your actual cost structure."

This frames the conversation around drivers, not discounts. The supplier either accepts the range or teaches you which assumption is wrong — both outcomes are useful.

Tail spend under ~$50k/year: the model takes longer to build than the savings pay back. Use published rate cards or reverse auctions instead.

Innovation partnerships and sole-source R&D: aggressive cost pressure kills the collaboration. Use open-book costing with a shared savings clause.

Regulated services (legal, audit, medical devices): the constraint is compliance, not cost — should-cost anchors are misleading.

How accurate does it need to be? Within ±10% of the true cost is enough to negotiate credibly. Chasing more precision rarely improves the outcome.

Do I share the model with the supplier? Share the target and the top 3 drivers. Never share the full breakdown — it removes their incentive to volunteer information.

How often should I refresh? Every 6 months for volatile categories (metals, energy, plastics) and every 12 months for stable ones (services, packaging).

Who should build it? A cross-functional squad: a category lead, an engineer or ops person who knows the process, and a data-savvy analyst. Purchasing alone doesn't have the technical depth.