On this page
Key takeaways
- Procurement analytics is the use of spend, supplier, contract and market data to answer questions that improve buying decisions.
- There are four types: descriptive, diagnostic, predictive and prescriptive. Most teams are still working to get the first two reliable.
- Fix data before you buy a tool. A single supplier list and a consistent category scheme matter more than any dashboard.
- McKinsey's 2025 survey found three quarters of companies were planning, blueprinting or piloting AI use cases, but only 19 percent were deploying AI tools at scale.
- Start with two or three questions the business already asks. Answer them well, then widen.
Free PDF: Procurement analytics readiness checklist. Get it below
A procurement analyst is asked a simple question by the finance director: how much did we spend with our top twenty suppliers last year? It takes her four days. The same supplier appears under three names, two categories overlap and one region reports in a different currency. When she delivers the answer, the director asks a follow-up that takes another four days. By then the budget meeting has passed. The analyst is capable. The data is the problem, and no tool will solve that on its own.
The scene is invented, and any analyst will recognize it. Getting out of it is the first job of procurement analytics.
Procurement analytics is the practice of collecting and analyzing data on spend, suppliers, contracts, prices and markets to give buyers evidence for decisions, to track results and to spot risk and opportunity earlier.
How does analytics differ from spend analysis and KPIs?
Our spend analysis guide covers classifying and cleaning spend, which is a foundation for analytics and one of its first uses. The procurement KPIs guide covers which measures to track. Analytics is the wider capability: the data, the skills and the methods that answer questions about all of it, including forecasting, supplier risk signals and price benchmarking. This guide covers that capability.
What are the four types of analytics?
| Type | Question | Procurement example | Data and skill needed |
|---|---|---|---|
| Descriptive | What happened? | Spend by category and supplier last year | Clean spend data, a basic report |
| Diagnostic | Why did it happen? | Why did price per unit rise in packaging? | Price and volume history, drill down |
| Predictive | What is likely to happen? | Which suppliers are likely to miss delivery next quarter? | History, external signals, statistical skill |
| Prescriptive | What should we do? | Which suppliers to keep and what volume to allocate | Models, rules and trusted data |
Teams are drawn to the last two. The first two do most of the work and are the base for the rest. A predictive model fed with unreliable history gives unreliable forecasts.
What data foundations do you need?
- One supplier list. One record per supplier, with parent company links. Duplicates break every report.
- A consistent category scheme. Use one classification across systems and regions. Many teams adopt a standard scheme, such as UNSPSC, then extend it where needed.
- Linked transactions. Purchase order, receipt and invoice connected, with the contract reference.
- A contract repository with dates, values, terms and owners.
- Currency and unit rules. Agree conversion dates and units.
- Ownership. A named person responsible for data quality, with a monthly check.
The Hackett Group projects procurement workloads to rise 8 percent in 2026 while headcount and budgets decline. Data cleaning is the first task to suffer. Protect time for it, because every later benefit depends on it.
Which use cases pay back?
| Use case | Question answered | Data needed | Typical first output |
|---|---|---|---|
| Spend visibility | Where does the money go? | Payments, supplier and category data | Spend by category, supplier and unit |
| Price variance | Do we pay different prices for the same thing? | Item level price and volume | List of items with the largest price spread |
| Contract coverage | How much spend is under contract? | Spend and contract data | Share of spend with no contract |
| Supplier risk signals | Which suppliers are showing stress? | Delivery data, financial and news signals | Watchlist of suppliers |
| Demand forecast | What will we need next year? | Past volume, business plan | Forecast by category with ranges |
| Savings tracking | Did the saving reach the budget? | Baseline, new price, volume | Validated savings report |
Start with the first three. They need little more than clean data and they find money. The use case for savings tracking is covered with its validation steps in procurement cost reduction, and the forecast case in procurement planning.
Free PDF
Procurement analytics readiness checklist
A readiness checklist covering data foundations, skills, use case selection and AI readiness, with a scoring sheet and a first ninety day plan.
- Data foundation check with scoring
- Use case chooser
- AI readiness questions
- First 90 day plan
Where does AI fit?
McKinsey's 2025 survey of 100 companies found three quarters were planning, blueprinting or piloting AI use cases, and only 19 percent were deploying AI tools at scale. EcoVadis's 2026 barometer of 1,000 multinationals reported that 55 percent of buyers had moved from experimenting with AI to operational use, mainly for ESG analytics and risk screening. Gartner's 2026 survey of 101 CPOs found an AI productivity paradox: individuals get faster, but gains shrink when you measure the whole team.
Read these three together. Pilots are common, scaled use is not, and the gains depend on process and data. AI that reads contracts or classifies spend needs good input. Weak data produces confident wrong answers. A short readiness check, in the download below, tells you whether you are ready to try.
What tools do you need, by stage?
| Stage | Typical tools | When it fits |
|---|---|---|
| Starting | Spreadsheets and a shared supplier list | Small team, low volume, first clean-up |
| Building | A spend analysis tool or a data warehouse with a reporting layer | Several systems, regular reporting needed |
| Scaling | Integrated suite with analytics, workflow and contract data | Larger team, many categories, need for automated alerts |
| Advanced | Predictive models, risk feeds, AI assistants | Reliable data, an analyst with statistical skill, clear use case |
Choose tools after you know the questions. For a view of the software side, see our procurement automation guide, and our procurement software reviews for options, which cover the tools while this page covers the capability.
How do you build the capability in stages?
- Pick two or three business questions. Ask finance and category managers what they most need to know.
- Audit the data. Check suppliers, categories, links between orders and invoices. Count the gaps.
- Fix the foundation. Merge duplicate suppliers, apply the category scheme and set ownership.
- Produce the first reports. Spend visibility and price variance. Share them and ask what is wrong.
- Act on one finding. Take a price variance and renegotiate or consolidate. A result builds support.
- Add one capability a quarter. Contract coverage, then risk signals, then forecasts.
What goes wrong?
- Tool first, data later. A new dashboard on dirty data is a faster way to be wrong.
- Too many metrics. A report with sixty measures is read for none.
- No owner for data quality.
- Analysis with no action. Every report needs a decision attached.
- Skills gap. An analyst who can ask good questions of the data matters more than any platform.
Gartner found only 36 percent of CPOs were very confident they could redesign roles and processes around AI. Build analytics skills deliberately, and see procurement change management for adoption. For where analytics fits in the plan, read procurement strategy, and see the full set of numbers in our procurement trends report.
Sources
- McKinsey, supply chain risk survey: 100 companies, December 2025. Three quarters planning, blueprinting or piloting AI, 19 percent deploying at scale.
- EcoVadis, 2026 Barometer: 1,000 multinationals, 55 percent of buyers in operational AI use.
- Gartner, CPO survey on redesigning procurement for AI: 101 CPOs, January to February 2026.
- The Hackett Group, 2026 Procurement Key Issues Study: sample size not published on the pages we read.
- The analytics types, use cases and build steps are editorial summaries of common practice. The opening case is invented.
Frequently asked questions
What is procurement analytics?
It is the use of spend, supplier, contract, price and market data to give buyers evidence for decisions, track results and spot risk and opportunity earlier.
What are the types of procurement analytics?
The four types are descriptive (what happened), diagnostic (why), predictive (what is likely) and prescriptive (what to do). Most teams need the first two reliable before the others.
What data do you need for procurement analytics?
You need one supplier list, a consistent category scheme, linked purchase order, receipt and invoice data, a contract repository, currency and unit rules and a named owner for data quality.
Which procurement analytics use cases give the quickest return?
Spend visibility, price variance and contract coverage need little more than clean data and often find money quickly. Add risk signals and forecasts later.
How should procurement use AI?
Start with a clear use case and reliable data. McKinsey found only 19 percent of companies deploying AI at scale, and Gartner found gains shrink at team level without redesign of roles and processes.