)
Where is your business spending its money, and where could it save? Spend analysis gives finance teams a clear view of company expenditure by bringing card, invoice, expense, and supplier data together. In this guide, we explain how to analyse spend in six practical steps, uncover savings opportunities, and improve financial control.
Key takeaways
Spend analysis only works once you pull card, invoice, and expense data into one cleansed dataset before looking for savings.
Start with a specific goal, such as a category to renegotiate or a compliance gap to close, and choose your data sources accordingly.
Group spend by supplier and category, then add the cost-centre view to reveal the biggest savings opportunities.
Duplicate suppliers and misclassified transactions distort every finding, so cleansing deserves more time than the analysis itself.
Automating capture and categorisation at the point of spend keeps the report current instead of requiring a full rebuild every quarter.
What is spend analysis?
When card, invoice, and expense data sits across separate systems, finance teams cannot reliably see where money goes or which savings opportunities deserve attention.
Spend analysis creates that reliable view by collecting, cleansing, classifying, and analysing expenditure. The Chartered Institute of Procurement & Supply defines it in its CIPS guidance as:
“The process of collection, classifying and analysing expenditure data,” run on historical spend “to provide answers to questions concerning spend visibility, compliance and control.”
For a finance team at a 50-to-1,500-person European company, that data may cover card transactions, supplier invoices, expense claims, and subscriptions across several entities and currencies.
The output is a spend report you can act on. It identifies contracts to renegotiate and suppliers to consolidate, while also showing purchases that bypassed approval.
Why create spend reports?
Spend reports turn expenditure data into decisions about supplier contracts and spending that has slipped past policy.
Without one, answering the question “Where is our money going right now?” usually requires finance to combine a card statement with an accounts payable export and then compare the result with a budget spreadsheet. By the time the answer is available, several days may have passed.
Most finance leaders know this gap exists. Efficio’s 2025 survey of 300 chief procurement officers and CFOs across the UK, Germany, the Nordics, and France found that only 19% were fully confident they had an accurate picture of indirect spend.
In the same survey:
85% said more than a quarter of indirect spend had no financial oversight.
93% named maverick spending as a major contributor to cost leakage.
If more than a quarter of indirect spend sits outside oversight, the variance analysis in your board pack is built on only part of the picture.
A spend report closes the gap by comparing the budget with approved amounts. It also shows what finance paid, bringing together information that would normally live in three separate systems.
Beyond visibility, a spend report gives you four practical benefits:
A negotiating position: Knowing your total volume with a supplier across every entity and cost centre is the starting point for a renegotiation.
Earlier warning on compliance: Off-contract purchases and duplicate payments become visible. Missing approvals surface too, allowing you to treat each issue as a process gap to close before an auditor finds it.
A better budget baseline: Next year’s budget starts from what was actually spent, by category and owner, instead of last year’s plan plus inflation.
Feedback for your controls: If the same category keeps landing off-contract, the fix belongs in the spend management system at the point of purchase, such as by adding a preferred supplier or setting a card limit.
How to do spend analysis in six steps
Spend analysis runs in six steps:
Set a goal.
Locate your spend data.
Consolidate and cleanse it.
Categorise and group it.
Analyse it against contracts and budget.
Build and share the report.
The order matters. A goal narrows the data you need, while cleansing is what makes the analysis in step five trustworthy. It also consumes most of the hours on the project.
Step 1: Set your goals
Decide the question the analysis must answer before exporting anything. Your goal determines which data you need and how finely you should classify it.
For example:
A project to cut software subscription spend needs card and subscription data at supplier level.
A project to prove approval compliance to an auditor needs purchase requests, approvals, and invoices linked together.
Common goals for mid-market finance teams include:
Cut spend in one category, such as software, travel, or agencies, by a set percentage.
Reduce supplier count and move volume to preferred suppliers.
Find spend that bypassed purchase orders or existing contracts.
Rebase next year’s budget on categorised actuals.
Document who buys what, so purchasing and procurement responsibilities are clear.
Every goal needs an attached metric. You can only show that a saving landed in the P&L if you have a baseline to measure it against.
For example, “Reduce maverick spend” becomes:
“Reduce off-contract spend as a share of total category spend, measured this quarter and again in two quarters.”
Step 2: Find out where your spend data lives
Spend data usually sits across four core systems:
Enterprise resource planning or accounting systems.
Accounts payable and invoice data.
Purchasing cards.
Travel and expense systems.
Purchase orders, contract repositories, and the supplier master usually come next.
Relevant accounts payable sources may include modules or separate invoice-processing tools.
Card data may come from:
Virtual cards.
Personal cards that employees use and reclaim.
Purchase orders may sit in procurement tools.
Map each source before requesting an export.
Source | What it gives you | What it is missing |
|---|---|---|
Accounting system or ERP | General ledger totals by account and entity | Supplier and category detail in a form you can group |
Accounts payable | Supplier, amount, date, and sometimes a purchase order reference, whether invoices sit in an AP module or separate processing tools | A purchase order reference on many lines |
Company cards | Statements from prepaid physical cards, virtual cards, and personal cards employees use and reclaim | A category and a cost centre |
Expense claims | Out-of-pocket spend and per diems | Structure of any kind. This is the source most likely to hide subscriptions |
Purchase orders and procurement tools | Committed spend that has not yet been invoiced, which the ledger does not show | Final amounts, which arrive with the invoice |
Contracts and supplier master | Agreed prices, payment terms, and legal entity names to check actuals against | What was paid |
Two sources can be especially difficult to identify:
Subscriptions paid on a founder’s or manager’s personal card.
Agency spend paid by a regional entity in another currency.
If your cards, expense claims, and invoices already run through a centralised spending platform, this step is largely an export. Employees attach the supplier and requester when the money moves, and they assign the cost centre at the same time.
Step 3: Pull everything together
Consolidate every source into one table with one row per transaction and a common set of fields.
The fields that matter for later grouping are:
Date.
Supplier or legal entity.
Amount.
Currency.
Entity.
Cost centre.
Category, where known.
Payment method.
Source system.
Keep the source-system column. It tells you later where duplicates came from.
Cleanse the data
Cleansing is where most of the hours go. Work through it in this order:
Normalise supplier names so that “AWS”, “Amazon Web Services EMEA SARL”, and “Amazon Web Svcs” become one supplier.
Convert every amount to your reporting currency on a stated rate basis, and record the rate used.
Remove duplicates, including the same invoice appearing in accounts payable and on a card statement, or an expense reimbursed twice.
Fill missing cost centres and categories where the source allows, and flag the rest instead of guessing.
Reconcile the consolidated total back to the general ledger for the same period.
Accounts payable automation can detect duplicate invoices, removing one of the cleansing passes.
The reconciliation is the step most often skipped. A spend report that does not tie to the ledger will not survive the first question from a board member or auditor. The untied difference is usually a source you forgot to map.
Step 4: Categorise and group spending
Assign every transaction to one category in a single taxonomy. Then group the result by supplier and category.
Add a cost-centre view to show who you buy it for. Together, these views answer:
Who do you buy from?
What do you buy?
Which business areas are responsible for the spend?
These dimensions form the axes of the spend cube.
For most mid-market companies, the right taxonomy is a custom hierarchy that mirrors your chart of accounts and budget lines, covering both direct and indirect spend.
Two or three levels are usually enough. Classification accuracy falls away below the second level, so a deeper tree may create more precise labels but also place more transactions in the wrong branch.
Categorising a year of card and invoice lines by hand is slow and error-prone. That is why rules do most of the work in a well-run process.
For example:
“This supplier always maps to this category and cost centre.”
Rules can handle recurring suppliers, while only new or ambiguous lines need human review.
Bookkeeping automation works on the same principle:
Deterministic rules set by finance handle known cases.
Machine-learning-based suggestions fill the gaps.
Finance reviews and confirms each prepared field instead of entering it manually.
Step 5: Analyse
With a clean, categorised dataset, the analysis compares:
Spend by supplier with contracts.
Spend by category with budget.
Payment behaviour with agreed terms.
Transactions across sources to identify duplicates.
CIPS breaks the work into four activities:
Consolidation and error elimination.
Commodity analysis.
Supplier analysis.
Contract-spend analysis.
Consolidation happens during cleansing. Finance teams identify savings through the other three activities.
Questions to ask of the spend cube
Which ten suppliers take the largest share of spend, and how many other suppliers sell you the same category?
How much of each category has a contract behind it, and how much is maverick spend bought outside preferred suppliers?
Which cost centres overshot budget in which categories, and did the relevant budget owners approve the overshoot before or after the fact?
What share of invoices matched a purchase order, and how many were exceptions that needed a manual fix?
Where are the duplicates, such as the same supplier invoiced twice or the same subscription running on two cards?
Which small, frequent purchases could move to a preferred supplier or a virtual card with a limit?
Compute a small set of metrics from those answers so that next quarter’s report can show movement.
Metric | What it measures |
|---|---|
Spend under management | Share of addressable spend that went to preferred suppliers under contract |
Maverick spend rate | Off-contract spend as a share of total spend, by category |
Invoice-to-purchase-order match rate | Share of supplier invoices linked to a purchase order |
Supplier concentration | Share of total spend with your top five or ten suppliers |
Realised savings | New cost against previous cost for the same volume, measured after the change |
On-time payment rate | Share of invoices paid within agreed terms |
Rank what you find by value and effort.
Recovering a duplicate payment may be a phone call. Consolidating six agencies into two may be a six-month project. Both belong in the report, but the owner, deadline, and expected saving will differ.
Step 6: Build and share your spend report
A useful spend report fits on a few pages and has four core parts:
Headline numbers.
Ranked opportunities with owners.
Compliance findings.
Data caveats.
Each part answers a different reader’s question, while the caveats protect the credibility of the other three.
Build the report in this order
Headline view: Total spend by category and entity against budget, with the same period last year alongside.
Supplier view: Top suppliers, concentration, and categories with too many suppliers.
Opportunities: Each with an estimated value, an owner, and a date, ordered by value against effort.
Compliance findings: Off-contract spend and duplicate payments, plus missing approvals, framed as process gaps to close.
Caveats: Share of spend left unclassified, currency basis, and any source excluded.
Different readers need different cuts:
The board and CFO need the headline view and the savings target.
Budget owners need their own cost centre because they hold the supplier relationships and will conduct the renegotiating.
Procurement teams need the supplier view, where they exist.
Send budget owners their slice with the opportunities they own, and put a review date in the calendar. A report without a follow-up date is only a snapshot.
If the categorised actuals feed into better business budgets for the next cycle, finance and budget owners can record the savings target as a budget line.
Spend analysis tips
Four practices make spend analysis faster and more reliable:
Build a spend cube.
Judge suppliers on total cost rather than unit price.
Route every invoice through one channel.
Measure your data quality before presenting findings.
Caliba Group: Try using spend cubes
A spend cube groups spend by supplier and category, then adds the cost-centre dimension.
Slicing it is one of the fastest ways to find consolidation and compliance opportunities.
One definition attributed to Caliba Group, a Brisbane procurement consultancy founded in 2011, describes it as:
“A unique way of taking a look at spend data in three dimensions: by suppliers, by business unit and by category of item or service.”
Slicing the cube surfaces patterns quickly:
A category bought from many suppliers across cost centres points to a consolidation opportunity.
A cost centre buying a category from a non-preferred supplier points to maverick spend.
One supplier charging different prices to different entities points to a contract that is not being applied group-wide.
A spend cube only shows money already spent. Open purchase orders and funded projects that suppliers have not yet invoiced sit outside it, so keep committed spend alongside the cube.
Zycus: Cut bait with poor vendors
Underperformance usually shows up in total cost long before it appears in a unit price.
A Zycus article, published by the procurement software vendor in April 2019 and last updated in October 2023, warns that vendors may promise the lowest cost while hiding overheads. It argues that a sourcing platform should:
“Analyse all your suppliers’ overall performance in the past, set performance benchmarks, forecast savings, and suggest the best performing supplier.”
For spend analysis, that means adding a total-cost view alongside unit price.
The comparison should include:
Late deliveries.
Rework.
Credit notes.
Finance time spent chasing a supplier.
Other operational costs.
A supplier whose landed cost keeps exceeding its quote is the one to benchmark against alternatives first. Finance can use the cube to calculate how much volume is at stake if the company moves it.
Tradogram: Designate a single point of entry for invoices
Route every supplier invoice through one channel and one owner. This makes accounts payable data usable for analysis without requiring a major cleansing project.
Tradogram’s guidance, written by co-founder and COO Majdi Sleimen, makes this its first recommendation:
“Designate purchasing roles and a single point of entry for invoices.”
This approach has several benefits:
One designated processing party can reconcile received invoices against purchase orders quickly, speeding up supplier payments.
Internal control improves when a three-way match is verified before details are passed to accounts payable.
A single route confirms that the purchasing team placed an order for each invoice the supplier sends, helping guard against fraud.
Invoices are less likely to sit unprocessed when someone is away.
Duplicate payment risk is reduced because a supplier cannot have the same invoice entered twice through different routes.
Some companies go further with a “no PO, no payment” policy.
An invoice that arrives by email and through a supplier portal may also arrive as a paper copy. That creates three records for one payment, leaving the cleansing stage to determine which one is genuine.
Cloudia: Bad data hurts just as much as no data
A spend report built on duplicated suppliers and misclassified lines produces confident conclusions that are wrong.
A supplier master that holds the same company under three legal names inflates the apparent number of suppliers serving a category. It can also distort the top-ten supplier list, meaning any consolidation target based on it points at the wrong problem.
Hand-classified lines carry an error band wide enough to move that target on their own.
It is worth sample-checking a few hundred classified lines against source documents before presenting the report. A wrong consolidation target costs more than an admitted data gap.
Report the following as early findings:
The unclassified share.
The duplication rate.
Any significant source gaps.
A board that sees the unclassified share on page one knows how much weight to put on the rest of the analysis.
Automate the tedious parts of spend analysis
Automate capture and supplier matching, then use rules for categorisation and duplicate detection.
These steps take most of the hours and are where duplicate supplier records and misclassified lines tend to creep in.
The remaining work, deciding which supplier to renegotiate with and which control to add, requires judgement. Automation should inform those decisions rather than replace them.
A peer-reviewed study of Cranswick plc, published in the INFORMS Journal on Applied Analytics in January 2025, estimated annual savings of:
3.27% to 3.43% of total procurement costs through automated spend analysis.
2.25% to 3.00% through manual analysis.
The study was modelled on one UK food producer, so the range should be treated as directional.
Researchers calibrated the simulation on two years of the company’s data, covering £1.571 billion of invoices across 2,170 suppliers.
The mechanism is what transfers:
Automated classification covers more of the spend.
Continuous analysis catches renewals and price rises when they happen.
Finance sees these issues months before an annual review would identify them.
The cheapest place to automate is the point of spend.
When an employee pays with a virtual card tied to a cost centre, or an invoice is captured with the supplier and category attached, consolidation shrinks from a cleansing project to a validation check.
For example, Spendesk is an all-in-one spend management platform that consolidates:
Company cards.
Expense management.
Accounts payable.
Procurement.
Budgeting.
Capturing requester and coding context with each transaction means finance can build the consolidated dataset as spending occurs instead of reconstructing it later.
An accurate picture of indirect spend comes from a dataset that reconciles to the ledger and refreshes without a quarterly rebuild.
The six steps produce that dataset once. Capturing data at source keeps it current for the next board pack.
For a visual explanation of how source capture supports spend reporting, get a free tour.
Frequently asked questions about spend analysis
These questions cover how often to refresh the analysis, what level of savings to expect, how long the work may take, and when a spreadsheet is sufficient.
How often should you run spend analysis?
Run a full analysis at least once a year, timed before budgeting, and refresh the core metrics at month-end.
An annual-only cycle means a price increase or auto-renewal in month two may go unchallenged for ten months.
A monthly refresh is realistic only when transactions are captured with the supplier, category, and cost centre attached. If classification is manual, a quarterly refresh of maverick spend, duplicates, and supplier concentration is a more honest target.
How much can spend analysis save?
Savings usually land in the low single digits, but the denominator and scope change the result.
APQC benchmarks put median cost takeout at 2.0% of annual purchased-spend value across 2,431 organisations.
Separate APQC measures put:
Cost takeout plus cost avoidance at 3.5% of all materials and services purchased.
Savings on sourcing events at 5.0% of sourcing-event spend.
These are broader procurement benchmarks rather than savings attributable only to spend analysis. The samples also skew towards larger organisations, so use them as directional baselines rather than guaranteed results.
How long does spend analysis take?
The timetable depends on:
The number of source systems.
The number of entities.
The number of currencies.
The volume of unclassified transactions.
The quality of supplier records.
A company with clean supplier records and centralised card and invoice data may complete its first report in several weeks.
Fragmented systems and inconsistent supplier names extend the work because cleansing and reconciliation must be completed before the findings are reliable.
Can you run spend analysis in a spreadsheet?
Yes, for a limited dataset with a manageable number of sources and categories.
A spreadsheet becomes harder to govern when you have:
Several entities.
Multiple currencies.
Supplier-name variants.
Recurring reporting cycles.
Large transaction volumes.
At that point, automated capture and classification reduce the risk that each reporting cycle becomes another manual rebuild.
Curious how Spendesk works?
Try an interactive demo to see spend control and approvals end-to-end.
Get a free tour)
Maxime Reding
Originally from Luxembourg and educated at the Rotterdam School of Management, he brings a sharp analytical mindset and a background in strategic management to our finance team. Maxime joined Spendesk in 2021 and has grown steadily through the ranks since, becoming our resident finance engineer.
)
)
)
)
)