Procurement KPIs that actually matter

Published 14 August 2026 · 5 min read

Most procurement scorecards measure activity rather than outcome — number of tenders run, POs processed, suppliers onboarded. Those are workload metrics. They tell you the team is busy, not that the organisation is better off.

A short list that actually informs decisions:

1. Spend under management

The share of total third-party spend that a procurement professional actively manages — contracted, sourced, with an owner. Not spend procurement is aware of; spend procurement controls.

This is the headline number because everything else is a subset of it. Below 60% and the function is reactive.

2. Off-contract spend by category

The percentage of category spend going to non-contracted suppliers. Directly actionable, unlike an aggregate compliance figure. See maverick spend.

3. Supplier concentration

What share of category spend sits with the top three suppliers. Interpret it in both directions: too low means no leverage, too high means dependency. There is no universal target — it depends on criticality and switching cost.

4. Classified spend percentage

How much of your spend is categorised at all. Rarely tracked, and it governs whether any of the above can be calculated. If 30% of spend is uncategorised, every category number you report is understated by an unknown amount.

5. Realised vs reported savings

Reported savings are what procurement claims. Realised savings are what finance can see in the ledger twelve months later. The gap between them is the most useful number on the scorecard and the least often published.

6. Contract coverage and expiry runway

What share of spend has a current contract, and how much expires in the next six months. The second half is the operational one — it tells you what to work on.

7. Payment terms, weighted by spend

Average days weighted by value, not a simple mean. Working-capital relevant and often the easiest concession to win in a negotiation where price is fixed.

What to leave off

Number of POs processed. Workload, not outcome.
Cost savings as a single aggregate. Mixes hard savings with cost avoidance and loses credibility.
Supplier count on its own. Fewer is not automatically better.
Tenders run. Measures activity, and can incentivise unnecessary sourcing events.

The prerequisite

Four of the seven require classified spend. That is why classification is not really an analytics exercise — it is the measurement infrastructure the rest of the function reports on. Consistency between periods matters more than granularity: a taxonomy that shifts each time you rebuild it makes every trend meaningless.