Should-cost modelling: a practical guide for buyers
When should-cost is worth the effort
- High-volume, high-spend components where a few percent moves the number
- Bespoke items with no market benchmark
- Categories with a small supplier base and limited competitive tension
- Any sole-source or single-source renegotiation
- New product introduction where target costing is required
The cost driver tree
Every should-cost model breaks price into five buckets:
- Direct material - specification, grade, yield, scrap and indexed input prices.
- Direct labour - cycle time, labour rate for the geography, indirect labour ratio.
- Machine / process - machine rate, setup time, tooling amortisation.
- Overhead - factory overhead, SG&A, freight, packaging, duty.
- Margin - a defensible profit assumption for the supplier's segment.
Where the data comes from
- Public commodity indices (LME, ICIS, Argus, Eurostat) for raw materials
- Regional labour rate databases and manufacturing benchmarks
- Reverse-engineered bills of materials and teardown analysis
- Supplier RFI data, filed accounts and analyst reports
- Internal engineering estimates - the most under-used source
From model to lower price
The model does not negotiate for you. Three moves make it pay:
- Anchor - open with a target price grounded in the model, not the current price minus 5%.
- Decompose - discuss line items (material, labour, overhead) not the total. It changes what "reasonable" looks like.
- Index - agree pass-through mechanics tied to public indices so future moves are automatic, not negotiated.
Common failure modes
- Model built once, never refreshed as indices move
- Engineering builds the model, procurement does not use it in the room
- Margin assumption too aggressive - supplier walks or quality slips
- No fallback if the supplier refuses to open the books
Related reading
Should-cost is one lever in a wider set. See the seven-lever cost reduction playbook and the sequenced strategy guide for how to combine it with demand and specification work.
What is should-cost modelling?
Should-cost modelling estimates what a product, component or service should cost to produce and deliver - built up from material, labour, machine, overhead and a fair margin - so buyers can negotiate against a fact base rather than last year's price.
When is should-cost worth the effort?
Any sole- or single-source renegotiation, high-volume components where a few percent moves the number, bespoke items with no market benchmark, and new product introductions where target costing is required.
How accurate does the model need to be?
Directional. A model within 10-15% of actual cost is enough to change the shape of a negotiation. Chasing the last 2% of accuracy usually costs more in engineering time than it recovers in savings.
What data sources feed a should-cost model?
Public commodity indices (LME, ICIS, Argus), regional labour rate benchmarks, teardown analysis, supplier RFI data, filed accounts and - most under-used - internal engineering estimates for cycle time and material use.
Related insights
More practical reading on procurement, cost and supply chain.
How a UK omnichannel retailer took 6.4% out of COGS in 14 weeks — packaging, freight, private-label sourcing and what made it stick.
Sequenced strategies for UK mid-market, SME, manufacturing and PE-backed businesses - with sizing and quick wins.
What indirect procurement covers, why it leaks, and how to build indirect procurement services that hold the savings.
Ready to find the savings hidden in your business?
Request a no-obligation Cost Review - typically 1–2 weeks, with minimal involvement from your team.