Process & tools

Spend analysis: how to find out what your organisation really buys

Process & tools · · Published 2026-10-05

Hands holding a white tablet showing a simple finance chart icon, beside a glass of coffee on a wooden café table
Reading figures on a tablet over coffee. Good spend analysis starts with clean supplier and category data, not with the chart.

Spend analysis answers three questions: what are we buying, from whom, and who is buying it? It sounds simple, but in most organisations the data is scattered across finance systems, purchasing tools, card statements and spreadsheets, with the same supplier spelt a dozen ways. Cleaning and classifying that data is the foundation for almost every other procurement activity.

Why it matters

Without a reliable picture of spend, a procurement team cannot prioritise. Spend analysis shows which categories are large enough to justify a sourcing project, where many small suppliers could be consolidated, where purchases bypass existing contracts, and where prices for the same item differ between sites. It also underpins risk work (which suppliers are critical?) and sustainability reporting: for many organisations, estimates of scope 3 emissions under the GHG Protocol Corporate Value Chain (Scope 3) Standard start from spend data multiplied by emission factors.

Step 1: Extract the data

Gather transactional data for a defined period, usually the last twelve months for a first exercise. Common sources:

If the organisation has several ERP systems or subsidiaries, extract from each and record the source so totals can be reconciled with the financial accounts. A spend cube that does not reconcile with the ledger will not be trusted.

Step 2: Cleanse and normalise suppliers

The same supplier often appears under many names ("ACME Ltd", "Acme Limited", "ACME UK", a subsidiary name). Normalisation groups these into one supplier, ideally linked to its ultimate parent, so you can see total exposure to a group. Useful techniques include standardising legal suffixes and punctuation, matching on tax or company registration numbers, and fuzzy matching on names and addresses with human review of uncertain cases. External identifiers such as the Legal Entity Identifier help for larger companies.

Step 3: Classify spend

Classification assigns each transaction to a category in a taxonomy: a hierarchy running from broad groups (IT, facilities, logistics, raw materials) down to specific categories (laptops, cleaning services, road freight, steel coil). Options include:

Classification can be based on the supplier (what does this supplier mainly sell?), the general ledger account, the line description, or a combination. Supplier-based rules quickly classify most spend because a large share typically sits with relatively few suppliers; line descriptions are needed for diversified suppliers such as distributors. Rule-based and machine-learning tools speed this up, but plan for human review of large or ambiguous items.

Step 4: Build the spend cube

The cleaned data can be viewed along three axes, often called the spend cube: category, supplier and business unit (cost centre, site, country). With it you can answer questions such as "how much do we spend on road freight, with which carriers, from which sites?" A pivot table can do this for a first pass; dedicated spend analytics tools add automation, refresh and dashboards (see the e-procurement software landscape).

Step 5: Turn analysis into opportunities

Common patterns to look for:

Prioritise opportunities by addressable value, ease of implementation and risk, and feed them into a category plan (see category management).

Keeping it alive

A one-off spend analysis is useful; a refreshed one is far more so. Improve data at source by requiring categories on purchase requisitions, maintaining clean supplier master data with a single creation process, and pushing more spend through purchase orders. Refresh the cube monthly or quarterly so savings and compliance can be tracked over time.

Common pitfalls