Direct vs indirect spend: a practical guide

Last updated: 2026-08-14

Published 14 August 2026 · 5 min read

Direct spend goes into the product you sell. Indirect spend keeps the organisation running. That is the whole definition — but the boundary is fuzzier than it sounds, and where you draw it changes how your taxonomy should be built.

The clean cases

DirectIndirect
Raw materialsOffice supplies
Components and sub-assembliesIT hardware and software
Primary packagingProfessional services
Contract manufacturingFacilities and utilities
IngredientsTravel

The arguable cases

Where teams disagree, and it usually does not matter as much as the argument suggests — provided you pick one and stay consistent:

  • MRO. Spare parts for production equipment. Not in the product, but production stops without them. Most manufacturers treat MRO as indirect and manage it like direct.
  • Production consumables. Gloves, lubricants, cleaning chemicals in a food plant. Not an ingredient; you cannot run without them.
  • Outbound logistics. Direct in some models, indirect in most.
  • Energy. Indirect for an office business, arguably direct for a smelter.

Why the split matters practically

The two behave differently enough that treating them the same causes problems:

  • Direct is fewer suppliers, larger values, specification-driven, and a change requires engineering or quality sign-off. Savings come from redesign and volume leverage.
  • Indirect is many suppliers, smaller values, high fragmentation, and buying is spread across people whose job is not procurement. Savings come from consolidation and compliance.

This is why indirect is where spend analysis usually pays off fastest. Direct spend is typically already visible — someone owns the bill of materials. Indirect is where the surprises live.

What this means for your taxonomy

A common mistake is building one flat taxonomy and forcing both through it. Direct spend generally wants depth aligned to your product structure. Indirect wants breadth aligned to functional categories.

A practical compromise: keep Level 1 universal so everything rolls up consistently for the CFO, then let Level 2 and Level 3 reflect what is actually in each side of the data rather than forcing symmetry.

Pick a rule and write it down

The cost of an imperfect boundary is small. The cost of an inconsistent one — MRO counted as direct this year and indirect next — is that no trend comparison works. Decide, document, move on.

Further reading


← All posts