What is spend analysis?

What does spend analysis mean?

Spend analysis is the process of gathering an organisation’s purchasing data from its finance and procurement systems, cleaning and classifying it, and analysing it to show how much is spent, on which goods and services, with which suppliers and by which parts of the organisation. Its purpose is to find where buying power can be used and where spending is out of control.

The US Government Accountability Office, in its 2004 report on spend analysis (GAO-04-870), described it as a tool that shows who the buyers and suppliers are, how much is spent on which goods and services, and where there are opportunities to leverage buying power. The report noted that leading private companies use it as the basis of a strategic approach to procurement, and recommended that agencies extract spending data automatically and repeatably from their existing systems.

Spend analysis in procurement is the starting point of strategic sourcing. The 2005 OMB memorandum on implementing strategic sourcing says an agency’s effort “begins with a spend analysis” and with identifying the commodities to source strategically. Without it, a sourcing plan rests on anecdotes about who the big suppliers are.

How do you do a spend analysis?

A spend analysis follows the same sequence whether it runs in a spreadsheet for one category or in dedicated software for a whole group. The first four steps are about data quality; experienced teams expect them to take most of the time.

1. Define the scope and period: which entities, which spend (indirect, direct or both) and usually the last twelve months, so that seasonal effects even out.
2. Extract the data: accounts payable invoice lines, purchase orders, purchasing card transactions, expense claims and, where relevant, contract data, from every system and entity in scope.
3. Cleanse and normalise suppliers: remove duplicates and reversals, convert currencies, and group the many spellings of one supplier, and its subsidiaries, under one parent name.
4. Classify each line into categories: map invoice lines or general ledger accounts to a taxonomy such as UNSPSC, CPV or an internal category tree, using descriptions, supplier type and accounts.
5. Enrich: add contract coverage, payment terms, supplier risk or diversity data, and the budget holder or business unit.
6. Analyse: build the spend cube, rank categories and suppliers, measure contract coverage and price dispersion, and identify the tail.
7. Act and refresh: turn findings into a sourcing pipeline and compliance actions, then repeat the extraction monthly or quarterly so the view stays current.

A one-off analysis is useful to start a programme, but the GAO’s point about repeatable extraction matters: savings are measured against a baseline, and a baseline that cannot be recalculated in the same way next year cannot prove anything.

What is a spend cube?

A spend cube is the structure most spend analyses produce: every amount spent sits at the intersection of three dimensions, so it can be sliced by any one of them or any combination. Spend cube analysis means reading those slices to find concentration, fragmentation and leakage.

Spend analysis dashboards and reports are views of the cube with filters. The most useful ones are few: spend by category with supplier counts, top suppliers with contract coverage, spend by business unit for a chosen category, and the tail, the long list of small suppliers that together take a small share of spend but a large share of invoices.

Which classification should a spend analysis use?

Classification decides what the analysis can show, so the taxonomy is chosen for the decisions it must support. Two public standards are common in procurement, and many organisations keep their own category tree on top of one of them.

Public bodies sometimes have their own codes too. In the US federal government, for example, OMB memorandum M-19-13 explains that addressable common spending, the spend suitable for category management, is identified using product service codes. Whatever the taxonomy, write down the mapping rules, because the classification of a line is a decision that someone will question when the savings are reported.

What does a spend analysis look like on real data?

The worked example below uses six invoice lines from a fictional company. It shows why supplier normalisation comes before any conclusion: six supplier names are in fact three suppliers, and the category picture changes once they are grouped.

Before normalisation, the largest single supplier appears to be Northline Logistics at €41,000. After it, Northline is €60,500 across two spellings, more than half of the spend, which makes freight the first category to examine: is there a contract, does it cover both entities that pay these invoices, and are the rates the same? The same logic applied to thousands of lines is what spend analysis software automates.

What questions does spend analysis answer?

The value of spend analysis is in the questions it lets a procurement team answer with evidence rather than opinion. The common ones are the same in most organisations.

• Category spend analysis: which categories are largest, how many suppliers each one has, and where volumes can be consolidated.
• Indirect spend analysis: how much goes to goods and services that support the business, such as IT, facilities, travel and professional services, which are often bought by many people outside procurement.
• Contract coverage: what share of spend in each category runs through a contract, and where the same item is bought at different prices.
• Tail-end spend analysis: how many suppliers make up the last few percent of spend, and whether catalogues, cards or a marketplace would handle them at lower process cost.
• Freight spend analysis and other specialist views: lanes, carriers, modes and surcharges, which need data beyond the invoice, such as weights and routes.
• Supplier risk: dependence on a single supplier or country in a critical category.

Spend analysis in supply chain management extends the same questions to direct materials: part prices across plants, the number of qualified sources per component and exposure to one region. The method is unchanged; the data comes from bills of materials and purchase orders more than from invoices.

Can spend analysis be automated with AI?

Automated spend analysis is mature for extraction, supplier normalisation and classification, the steps that consume most of the effort. Rules and machine-learning classifiers assign categories from invoice descriptions and supplier names, and AI spend analysis tools add language models that read free-text descriptions and suggest codes. The output still needs review: a classifier is only as good as the taxonomy and the corrections fed back to it.

Organisations choose between three set-ups: an analysis built in-house in spreadsheets or a business-intelligence tool, a spend analysis module in a source-to-pay suite, or spend analysis services and consulting, where an outside team cleans and classifies the data. The right one depends on how often the analysis must be refreshed and who will act on it. Whichever is chosen, keep the classification rules and the supplier mapping as your own data, so the baseline survives a change of tool or provider.

What makes a spend analysis useful?

• Start from the decisions it must support, such as next year’s sourcing pipeline, and size the effort to them.
• Agree the taxonomy with the category owners before classifying, so that results land in the categories they manage.
• Show data quality openly: the share of spend classified, the share with a known supplier parent, and the period covered.
• Refresh on a fixed schedule and with the same rules, so that savings and contract coverage can be compared over time.
• Turn each finding into an owner and a date in the category management plan; an analysis without actions is a report.