Expense report errors: causes, controls and KPIs

Not every anomaly in an expense report is fraud. Most are simply mistakes: an amount typed in the wrong currency, a meal booked to the wrong cost centre, a missing receipt, a hotel booked outside the preferred travel agency. Taken one by one, these errors look harmless. Accumulated across thousands of claims and dozens of entities, they degrade the reliability of accounting data, weaken the audit trail and, above all, create the background noise in which real risks can hide.

This article is part of a series that grew out of discussions with members of our User Club: finance, internal control and audit leaders from international groups operating across many countries and entities. Errors and poor practices were seen as a medium-priority family of risks, but one with a disproportionate effect on everything else.

Expense management platforms have removed much of the manual work. Receipts are scanned, amounts extracted automatically, policies partly embedded. And yet errors remain stubbornly frequent, and each one has a cost: time spent by the employee to correct the claim, by the approver to review it again and by the finance team to reprocess and adjust the accounts.

Five recurring issues stand out:

• data entry errors on amounts or currencies;
• incorrect categorisation or accounting allocation;
• misuse of corporate cards;
• missing or lost receipts;
• misapplication of the expense policy (caps, eligible expense types, use of the preferred travel agency).

Two factors explain their persistence in large groups. The first is diversity: multiple currencies, local charts of accounts, country-specific rules and several tools coexisting across entities. The second is tolerance. When the same error is corrected again and again without consequence, it stops being an error and becomes a practice. The line between a mistake and a habit is thin, and the line between a habit and abuse is not much thicker.

Most controls focus on the claim itself, inside the expense tool. But a significant share of errors arises at the interfaces: between the claim and the corporate card statement, and between the expense tool and the general ledger.

Between the tool and the ledger. An approved claim may never reach the accounts, or reach them with a different amount. Conversely, expenses may be booked directly in the ledger without ever passing through the expense tool, and therefore without approval or policy checks. Long delays between the date of the expense, its submission and its accounting entry blur the picture further.

Between the expense process and the purchasing process. Expense reports are sometimes used as a shortcut around procurement: software subscriptions, small services or recurring purchases claimed by employees rather than ordered through a purchase order. Each claim is modest, but together they escape supplier controls entirely.

Between the card and the claim. A hotel paid with a corporate card and also claimed as an out-of-pocket expense, or a personal purchase made on a corporate card and never justified, both create gaps that no single document reveals.

None of these issues is visible from inside the expense workflow alone. They only appear when expense data is compared with accounting, card and purchasing data

For each risk, a key control applies:

• Data entry errors: automatic checks on currencies, formats and thresholds.
• Incorrect allocation: a maintained analytical framework and mandatory drop-down lists.
• Corporate card misuse: regular controls on card usage and reconciliation of card transactions with claims.
• Missing receipts: automatic blocking of any claim without a supporting document.
• Policy misapplication: rules embedded in the tool (caps, categories, eligible expense types)

The common thread is to prevent errors at the source rather than correct them afterwards. But preventive controls need calibrating. Blocking everything is counterproductive: it frustrates employees, generates workarounds and buries reviewers in exceptions. The groups involved favour alert thresholds over systematic blocking, and reserve hard blocks for clear-cut errors such as a missing receipt.

One practice was singled out as structuring: maintaining a dedicated “expense reports” journal in the accounts, fed automatically from the expense tool.

The benefits go well beyond tidy bookkeeping. With a dedicated journal, any expense booked in travel or professional expense accounts outside that journal stands out immediately. That makes it possible to detect claims processed outside the tool, identify workarounds of the purchasing process and secure the audit trail from the receipt to the ledger.

This practice relies on systematic consistency checks between the tool and the accounts:

• approved claims not found in the ledger;
• expenses booked without a counterpart in the tool;
• differences between approved and recorded amounts;
• entries without document references;
• abnormal delays between expense, submission and posting.

An illustrative case shows the value of this approach. A group introduces a dedicated journal and runs a first comparison of its travel and expense accounts. It finds a series of entries booked manually in one subsidiary: recurring software subscriptions, paid by employees and reimbursed outside the expense tool. None had gone through approval or policy checks, and none had been seen by procurement. Individually immaterial, they represented a recurring flow that no one was looking at.

Zero errors is neither realistic nor a useful target. What matters is knowing the error rate, understanding where errors come from and checking that controls reduce them over time. Several indicators help:

• the reconciliation rate between the expense tool and the ledger, including amounts and timing gaps;
• the share of expenses booked outside the dedicated journal in travel and expense accounts;
• the rejection and send-back rate per employee and per entity, to locate recurring sources of error;
• the share of claims without a receipt or document reference;
• average delays between expense date, submission and accounting entry.

Tracked over several months, these indicators show whether errors are isolated or structural, whether some entities or populations concentrate them, and whether embedded controls actually have an effect. A high rejection rate concentrated on a few employees, for instance, may point to a training need, or to repeated testing of the system’s limits.

Expense errors are usually treated as an operational cost: time spent correcting, reimbursements delayed, accounts adjusted. They are also a signal. Recurring errors reveal gaps between policy and practice, between tools and accounts, between the expense process and the purchasing process. Left unaddressed, they create precisely the noise in which fraud can go unnoticed.

The groups that manage them best do not try to block everything. They embed simple rules in the tool, reconcile systematically with accounting, and measure over time whether errors decline.

The next article in this series looks at a family of risks often considered lower priority, but whose consequences can surface long after the fact: tax and regulatory non-compliance.