Join our Newsletter — 33% off our NHI Course
Home› Guides› Procurement and Expense Fraud Detection Guide: Controls and…
Guide Fraud Prevention

Procurement and Expense Fraud Detection Guide: Controls and Analytics for Purchase to Pay

PDF for newsletter subscribers · free to read online

← All guides
By NHI Mgmt Group, edited by Lalit Choda Updated 14 November 2026 14 min read
On this page

Fake vendors, split purchases and padded expense claims succeed because each transaction looks normal on its own, and only the pattern across many transactions gives them away. This guide is the control and analytics manual for the purchase to pay (P2P) cycle and for employee expenses. It sets out where fraud enters each stage, the preventive controls that close the gap, the detection tests that find what got through, and how to run those tests continuously without drowning investigators in false positives. It is written for finance controllers, internal auditors, counter-fraud teams and the identity and ERP teams who own the access that makes these schemes possible. For how occupational fraud is classified and how a fraud programme is governed, see the Occupational and Internal Fraud Guide.

Key takeaways

  • Control the vendor master first. Almost every billing scheme needs either a fake supplier or a changed bank account; independent verification of every new vendor and every bank detail change closes the largest door.
  • Thresholds attract fraud. Approval limits, tender thresholds and receipt limits create clusters of transactions just below them; test aggregate spend per supplier and per requester, not single transactions.
  • No purchase order, no pay. Invoices without a matching purchase order and goods receipt are where non-delivery, inflated invoices and personal purchases hide.
  • Expense fraud is volume fraud. Individual claims are small, so the controls that matter are automated: card matching, duplicate receipt detection and policy rules applied to every claim.
  • Access design is fraud prevention. A user who can create a vendor, approve an invoice and release payment can run a scheme alone; segregation of duties in the ERP is owned by the identity team as much as by finance.

Where fraud enters the purchase to pay cycle

Map schemes to the stage where they start, because that is where the preventive control belongs. The Northern Ireland Audit Office's Procurement Fraud Risk Guide organises its red flags the same way: pre-tender, tender, award, contract management and payment, and non-competitive purchasing.

StageTypical schemesPreventive controlDetective test
Need and specificationSpecification written around one supplier; unnecessary purchases; requirements split to avoid tenderDocumented needs assessment approved by someone other than the requester; procurement involvement above a set valueRepeat purchases of the same item from the same supplier in a short window
Supplier onboardingShell companies; employee-owned suppliers; supplier impersonationVendor creation by a master data team, never by buyers or AP clerks; registry and tax ID checks; bank detail verification by call-backVendor to employee matching on address, bank, phone, tax ID; vendors with only a PO box or residential address
Tender and awardBid rigging, cover bids, leaked bids, kickbacks, changed scoringEvaluation criteria fixed before tender; e-tendering with locked submissions; panel conflict declarationsSame winner pattern, low bidder counts, winning bidders subcontracting to losers
Ordering and receiptGoods never delivered; deliveries to home addresses; personal items on company POsGoods receipt by someone other than the requester; approved delivery address listPOs with delivery addresses not on the approved list; receipts entered by the PO creator
Invoice and paymentDuplicate invoices, inflated prices, fictitious invoices, changed bank detailsThree-way match (PO, goods receipt, invoice); dual release of payment runsDuplicate invoice tests; payments to recently changed bank accounts; price drift against contract
Contract managementUnjustified variations after a low bid; overbilling on time and materialsVariation approval outside the contract owner; rate cards in the systemContracts where spend exceeds award value; variation counts per contract owner

Vendor master controls that stop fake suppliers

The vendor master is the single most valuable data set for P2P fraud control. Treat it like a privileged system.

  • Separate creation from use: A central master data team creates and changes vendors. Buyers request, they do not create. AP clerks pay, they do not edit.
  • Verify identity at onboarding: Check the company registry (Companies House in the UK, state filings in the US), confirm the tax identifier, and in the US validate name and TIN combinations through the IRS TIN Matching programme, which lets authorised payers check them before filing information returns.
  • Verify bank details out of band: Call the supplier on a number from an independent source, never from the request. Use payee name checking services where your banking market offers them. See the Invoice Redirection and Supplier Bank Detail Fraud Guide for call-back design.
  • Log every field change: Bank account, address, payment terms and tax ID changes should each generate a log entry with the user, the old value and the new value, and should be reviewed by someone outside the team that made them.
  • Close dormant vendors: Block vendors with no activity for 12 to 18 months. Reactivated dormant vendors are a classic vehicle for fictitious invoices, and the NIAO lists closure of dormant accounts among its payment controls.
  • Collect conflict declarations: Require annual and per-procurement conflict of interest declarations from buyers, approvers and evaluation panels, and match them against the vendor master.

Split purchases and threshold avoidance

Every approval limit creates an incentive to stay just below it. In US federal procurement, FAR 13.003(c)(2) prohibits breaking down requirements that aggregate above the simplified acquisition threshold into smaller purchases merely to use simplified procedures or to avoid requirements that apply above the micro-purchase threshold. Most private sector delegation of authority policies carry an equivalent rule, but few test for it.

Test for splitting along three dimensions:

  1. Same requester, same supplier, short window: Two or more purchase orders or card transactions within a few days whose combined value exceeds the requester's limit.
  2. Same supplier, many requesters: Aggregate spend with one supplier over 12 months that exceeds the tender threshold without a tender, which the NIAO flags as a sign of splitting in non-competitive purchasing.
  3. Value clustering: A histogram of transaction values for each approval band. A spike in the last 5 to 10 percent below a threshold, compared with the rest of the band, points to deliberate avoidance.

Expect legitimate explanations, such as genuinely separate needs or standing orders. The test output is a list of questions for budget holders, not a list of fraudsters.

Bid rigging and tender manipulation signals

Collusion between bidders, and between bidders and staff, defeats competitive tendering while leaving a paper trail that looks compliant. The US Department of Justice Antitrust Division describes four common forms: bid suppression, complementary (cover) bidding, bid rotation, and subcontracting arrangements where losing bidders are paid through subcontracts.

SignalWhat it may indicateData to test it
The same supplier keeps winning, or suppliers appear to take turnsBid rotation or market allocationContract register with all bidders and outcomes over several years
Fewer bidders than usual; firms download tender documents but do not bidBid suppression or a restrictive specificationE-tendering platform access logs
Losing bids far above price lists or estimatesComplementary biddingBid prices against engineering estimates or benchmarks
Prices drop when a new bidder entersAn existing cartel being undercutBid price history by lot
Winner subcontracts to losing biddersCompensation for losingSubcontractor declarations against bidder lists
Scoring changed after bids opened, late bids acceptedInternal manipulation or briberyE-tendering audit trail, panel records

The DOJ notes these indicators are not proof of collusion. They justify a closer review and, where collusion is suspected, referral to the competition authority as well as internal investigation.

Expense and corporate card fraud

Expense schemes fall into four familiar patterns: mischaracterised expenses (personal costs claimed as business), overstated expenses, fictitious expenses and multiple reimbursement of the same cost. Corporate and purchasing cards add personal use and transactions at inappropriate merchants.

  • Card-first policy: Pay as much as possible on corporate cards so the claim is matched to bank-reported transaction data rather than an employee-supplied receipt.
  • Merchant category controls: Block merchant category codes for cash, gambling and other categories you never expect to pay, and set single transaction and monthly limits per card. The NIAO recommends limits on card numbers and users, spending caps and restricted categories.
  • Receipt integrity: Require itemised receipts above a low threshold, detect the same receipt image or same merchant, date and amount across claims, and treat edited or generated receipt images as a growing risk to test for.
  • Approval design: Approvers review against policy with the receipt image in view, cannot approve their own claims or claims for their own manager, and senior executive claims go to an independent reviewer.
  • Mileage and per diem: Compare claimed mileage against route distances and check per diems against travel bookings and calendar data where policy allows.
Expense testData neededTypical finding
Duplicate receipts across claimants or periodsReceipt metadata, merchant, date, amountTwo people claiming the same meal; resubmitted claims
Cash or out-of-pocket claims for costs normally paid by cardCard feed and expense claimsFictitious or duplicate costs
Weekend, holiday and leave-day expensesClaims, HR absence dataPersonal costs claimed as business
Claims just under receipt or approval limitsClaims by amount bandThreshold avoidance
Round amounts and repeated identical amountsClaims history per employeeEstimated or invented costs
Card transactions with no matching claim after the deadlineCard feed, claim statusPersonal use left unexplained

For expense platform configuration, such as audit rules, delegate settings and approval flows, see the SAP Concur Security guide.

Running analytics continuously

One-off audits find old fraud. Continuous testing shortens how long a scheme runs. Build the programme in layers.

  1. Connect the data: Vendor master with change log, purchase orders, goods receipts, invoices, payments, card feed, expense claims, HR master (addresses, bank accounts, leavers), and ERP user and role assignments.
  2. Start with high-precision tests: Duplicate payments, vendor to employee bank matches, payments to bank details changed in the last 30 days, and invoices without a purchase order. These produce few false positives and quick recoveries.
  3. Tune before widening: Run each new test in shadow mode for a month, review every hit with the process owner, record why each was or was not a concern, and adjust thresholds before alerts go to investigators.
  4. Add pattern tests: Split purchase detection, value clustering, supplier concentration per buyer, price drift after award and digit frequency analysis. Treat Benford's law results as a pointer to where to look, not as evidence.
  5. Score and route: Combine test hits into a risk score per vendor, buyer and employee so investigators work the highest combined risk first. See the Fraud Rules and Machine Learning Models Guide.
  6. Close the loop: Feed outcomes back into thresholds and preventive controls. A test that keeps finding the same gap means a preventive control is missing.

Rolling back is simple and should be planned: if a new test floods the queue, return it to shadow mode rather than raising thresholds blindly. See the Audit Evidence and Continuous Controls Monitoring Guide for evidence handling.

Where identity fits

P2P fraud depends on access. The toxic combinations to remove are well known: create or change vendor plus approve invoice; create purchase order plus enter goods receipt; approve invoice plus release payment; maintain employee bank details plus run payroll or expenses. Test both what users can do (role assignments) and what they actually did (transaction logs), because firefighter and emergency access often bypasses role design. See the Segregation of Duties Guide and the Segregation of Duties Matrix Template.

Service accounts and integration users that post invoices or payment files into the ERP need the same scrutiny: a compromised or misused integration credential can create vendors or change bank details at scale. See the ERP platform integration security guide.

Decision guidance: small and large organisations

  • Small organisations: Segregation is hard with a two-person finance team. Compensate with owner or director review of the vendor master change report and the payment run each cycle, bank-side dual authorisation, and corporate cards with low limits. Spreadsheet-based duplicate and vendor to employee tests run monthly are enough to start.
  • Mid-sized organisations: Introduce a master data role, three-way match in the ERP, and a quarterly analytics run owned by internal audit or finance control.
  • Large organisations and public bodies: Continuous monitoring with a dedicated analytics team, e-tendering for all competitive procurement, contract registers reviewed for bidder patterns, and data sharing with the counter-fraud function.
  • Build or buy: ERP-native controls cover three-way match and workflow; specialist analytics or continuous monitoring tools add cross-system tests. Buy when the volume of tests and data sources outgrows what an audit analytics team can maintain.

Detection and measurement

Measure both control health and detection outcomes.

  • Vendor master hygiene: Percentage of bank detail changes verified by independent call-back, and number of active vendors with no transactions in 12 months.
  • PO compliance: Share of invoice value paid with a matching purchase order and goods receipt.
  • Threshold clustering: Ratio of transactions in the 10 percent band below each approval limit to the band below that, tracked over time.
  • Test precision: Proportion of analytics hits that lead to a confirmed issue, error or recovery, per test.
  • Time to detect: Days from the first fraudulent transaction to detection, for confirmed cases.
  • Recoveries: Value of duplicate payments and overcharges recovered per quarter.

For a fuller metrics set, see the Fraud Metrics and KPIs Guide.

Framework mapping

Control areaNIST SP 800-53 Rev. 5NIST CSF 2.0ISO/IEC 27001:2022 Annex A
Segregation of duties in ERP and expensesAC-5 Separation of DutiesPR.AA-05A.5.3
Review of transaction and change logsAU-6 Audit Record Review, Analysis, and ReportingDE.AE-02A.8.15, A.8.16
Insider fraud as part of insider riskPM-12 Insider Threat ProgramGV.RR-02A.6.4
Supplier due diligenceSR-6 Supplier Assessments and ReviewsGV.SC-07A.5.19

At programme level, the GAO Fraud Risk Management Framework (GAO-15-593SP) organises fraud risk management into four components, commit, assess, design and implement, and evaluate and adapt, with an emphasis on prevention.

Common failure patterns

  • Buyers who can create vendors: Letting procurement staff add suppliers "to speed things up" removes the main barrier to shell company schemes.
  • Bank changes verified by email reply: Confirming a change with the same channel that requested it verifies nothing; the call-back must use an independent number.
  • Testing transactions, not aggregates: Single-transaction rules miss split purchases and suppliers that grow past the tender threshold without a tender.
  • Alert floods with no tuning: Launching dozens of tests at once buries investigators and teaches them to dismiss alerts.
  • Executives outside the controls: Senior staff claims approved by their own assistants, or exempt from receipt rules, are where large expense frauds run longest.
  • Ignoring emergency access: Firefighter IDs used to post vendor changes or payments, with no after-the-fact review, undo the segregation of duties design.

Practitioner checklist

  • Move vendor creation and change to a master data team and remove that access from buyers and AP.
  • Verify every new vendor against a registry and tax identifier, and every bank change by independent call-back.
  • Enforce no purchase order, no pay with three-way match, and approved delivery addresses.
  • Close dormant vendors and review the vendor master change log each payment cycle.
  • Run duplicate payment, vendor to employee and recent bank change tests first, then add split purchase and clustering tests.
  • Monitor aggregate spend per supplier against tender thresholds and review the contract register for bidder patterns.
  • Put cards first for expenses, block unwanted merchant categories, and detect duplicate receipts automatically.
  • Remove toxic access combinations in the ERP and review emergency access use.
  • Track test precision and time to detect, and feed outcomes back into preventive controls.

Standards and references

Related NHI Mgmt Group resources: Occupational and Internal Fraud Guide · Segregation of Duties Guide · Invoice Redirection and Supplier Bank Detail Fraud Guide · SAP Concur Security · Insider Threat Programme Guide

PDF for subscribers

Take this guide with you

Every section, table and checklist in one PDF, with links to the sources and related guides. Subscribe to our weekly NHI & AI Identity Journal and we’ll email you a download link.

Free weekly newsletter

Subscribe to the NHI & AI Identity Journal

The latest on NHI and Agentic AI security – articles, research, breaches, news and events every week.

Bonus 33% off our NHI Course when you subscribe.

NHIMG Editorial Note
Researched and written by the NHI Mgmt Group editorial team using AI tools, under the editorial direction of Lalit Choda. Facts are drawn from the sources listed on this page and checked by our automated quality process. Last updated 14 November 2026. Spotted an error? Use our contact page.
NHI Mgmt Group, the independent authority on Non-Human Identity, IAM and Agentic AI security. nhimg.org

PDF Downloads for Newsletter Subscribers

Get the guide

Enter your details and we’ll email you a download link.

The full guide is free to read online. The PDF is a thank you for subscribing.

Fill in all fields marked * and tick the PDF box to continue.