Building a spend cube for a machine builder: categories, tail spend and should-cost
A spend cube answers three questions: what we buy, from whom, and for which machines. How to build one from ERP data, a category tree that fits machine building, and where the savings usually sit.
Not ready to talk yet? Take the 3-minute Procurement Data Check
TL;DR
- A spend cube breaks spend down by category × supplier × business unit. Machine builders should add a fourth axis: machine type, via the BOM.
- Most of the work is cleansing: supplier names, categories, units and currencies.
- Tail spend rarely saves much per dollar but costs a lot of process effort. Consolidate it.
- Should-cost on the top drawing parts gives buyers facts instead of last year’s price.
What data goes in?
| Source | Content | SAP example |
|---|---|---|
| Purchase orders | Item, supplier, quantity, price, date | EKKO / EKPO |
| Invoices | What was actually paid | RBKP / RSEG |
| Supplier master | Names, groups, countries | LFA1 (Business Partner in S/4HANA) |
| Material master | Material group, description | MARA / MAKT |
| BOMs | Which machine uses which part | MAST / STKO / STPO |
Which category tree fits machine building?
| Level 1 | Level 2 examples |
|---|---|
| Machined & fabricated parts | Turned, milled, sheet metal, welded assemblies |
| Castings & forgings | Grey iron, aluminium, steel forgings |
| Drives & motion | Motors, gearboxes, linear guides, ball screws |
| Electrical & automation | Controls, drives, sensors, cabinets, cables |
| Fluid power | Hydraulics, pneumatics, valves, cylinders |
| Standard & C-parts | Fasteners, bearings, seals |
| Purchased systems | Robots, chillers, conveyors |
| Indirect | Services, logistics, IT, facilities |
Don’t use the material group as your category
ERP material groups were set up for posting and MRP, often decades ago. Map them to a procurement category tree, and categorise by description and supplier where a material group is too broad.
What does the cube show?
Sort suppliers by spend, descending
Top suppliers = suppliers covering the first 80% of spend
Tail suppliers = everyone else
Typical picture: a few suppliers carry 80% of spend,
many share the remaining 20%
Tail cost driver: number of POs, invoices and supplier records, not priceWhere are the savings levers?
| Lever | Where it applies | Data needed |
|---|---|---|
| Supplier consolidation | Fragmented categories, tail | Spend per supplier and category |
| Should-cost | Drawing parts, castings | Drawing, material, process times |
| Volume bundling | Same part from several suppliers | Deduplicated part master |
| Specification | Over-specified parts | Engineering review, standards |
| Terms & catalogues | Tail, C-parts | PO and invoice data |
Should-cost in four steps
- Pick the top 50 drawing parts by spend.
- Model material + machining time + set-up + overhead + margin.
- Compare with the price paid.
- Take gaps above roughly 15% into negotiation, with the cost breakdown on the table.
Checklist
- 24–36 months of POs and invoices
- Supplier names normalised and grouped
- Procurement category tree mapped to material groups
- Spend linked to machine types via the BOM
- Should-cost model for the top drawing parts
FAQ
What is a spend cube?
A data model that breaks purchasing spend down by category, supplier and business unit, usually over time. It shows where money goes and where the savings levers are.
Which data is needed to build a spend cube?
Purchase orders, invoices, the supplier master and the material master. For machine builders, BOMs add the link between spend and machine types.
What is tail spend?
The many small suppliers that together account for a small share of spend. They cost little in price terms but a lot in process effort.
What is should-cost analysis?
Calculating what a part should cost from material, process times, overheads and margin, and comparing it with the price paid. It gives buyers facts for negotiation.
How often should the spend cube be refreshed?
Monthly or quarterly from ERP exports. A one-off cube goes stale within a year.
What does your spend look like by machine type?
Send us 24 months of purchasing data under NDA. We’ll return a cleaned spend cube with your top savings levers.




