When Finance Teams Become Data Clean-Up Teams

Observable condition

Month-end starts, and the finance team begins correcting. Expenses have been coded to the wrong cost centre. A supplier invoice has been captured twice. A customer has two accounts because sales created a new one instead of finding the existing record. Sales orders reference products that were renamed. Job numbers on timesheets do not match job numbers in the accounting system. A credit note was raised without a reason code. None of this is accounting. All of it must be fixed before the accounts can be closed.

Finance staff become experts in other departments' mistakes. They know which branch always codes fuel to the wrong account and which salesperson types customer names differently. They keep correction spreadsheets and lookup tables that are updated every month. When asked what they spend their time on, they describe reconciliation, journals and queries, but much of it is data repair.

An illustrative example: a Pretoria-based engineering services firm with three branches has a finance team of four. Each month-end takes about eight working days. When the financial manager looked at where the time went, most of it was not producing the accounts. It was correcting job codes, matching supplier invoices to purchase orders that were never captured, and merging duplicate customer accounts before the debtors' report could be trusted.

Realisation

When finance becomes a data clean-up team, the business has placed its quality control at the end of the process instead of the beginning. Errors are created cheaply, in the moment of capture, by people who often do not know they have made them. They are then corrected expensively, weeks later, by people who did not make them and have to investigate what was meant.

Data quality writing makes the same distinction. Data cleansing corrects inaccurate or incomplete records after they exist. Data validation rejects invalid data at the time of entry. Validation is cheaper because the person who knows the right answer is present when the error occurs. A finance team that spends month-end cleansing is doing, late and indirectly, the validation that the systems upstream never performed.

There is also a compliance dimension. South Africa's Protection of Personal Information Act requires responsible parties to take reasonably practicable steps to ensure that personal information is complete, accurate, not misleading and updated where necessary. Duplicate and inconsistent customer records are not only an accounting nuisance.

Realisation prompt: what share of your finance team's month-end is spent producing numbers, and what share is spent repairing data so that numbers can be produced?

Identification

This is a data-quality-at-source problem. The visible symptom is a long, correction-heavy month-end. The common assumption is that finance is under-resourced or that the close process needs tightening. The structural cause is that operational systems and processes allow incomplete, inconsistent and duplicate data to be captured, and finance is the first function that cannot proceed until it is fixed.

It overlaps with, but differs from, the cost-focused Impact pieces on manual month-end close and manual reconciliation between systems. This Pattern is about where the errors originate and why finance ends up owning them.

Diagnosis

IBM describes data quality in terms of dimensions such as completeness, uniqueness, validity, timeliness, accuracy and consistency. Finance clean-up work maps onto these closely.

Behind these sit organisational causes. Nobody outside finance is measured on data quality. Errors are corrected silently rather than returned to the person who made them, so the same errors recur. Finance staff build private correction tables instead of changing the source system. Ownership of master data, such as customers, suppliers, products and cost centres, is unclear, so anyone can create records. The Pattern Why Does the Same Customer Have Three Different Names in Your Systems? looks at one of these causes in detail.

Commercial impact

The commercial value at stake is capacity, speed, cash flow, risk and decision quality.

Capacity is lost because skilled finance staff spend their time on repair rather than analysis, cost control or cash management. Speed suffers because a long close delays management information, so decisions are made on last month's figures well into this month. Cash flow is affected when customer duplicates and missing references slow invoicing, statements and collections. Risk increases because corrections made under time pressure can introduce new errors, and because inconsistent personal information can create compliance exposure. Decision quality falls when managers do not trust the numbers and start keeping their own.

The problem also limits automation. Clean-up that is invisible when people do it becomes visible failure when software does, as described in When Poor Data Quality Becomes an Automation Failure.

Common misidentification

Adding finance staff or buying a faster close tool are common responses. Both treat the symptom. More people can clean more data, and a close tool can organise the tasks, but neither stops the errors arriving.

Another misidentification is blaming individuals in other departments. The people capturing data are usually doing what the system allows under time pressure. If the system permits a wrong code, a missing reference or a duplicate record, it will keep receiving them.

Possibility

In a better state, the systems where transactions start prevent the most common errors. Required references must be present. Cost centres and account codes are defaulted or restricted by rule. Searching for an existing customer or supplier is required before creating one, and creation is limited to a few people. Operational and finance systems share identifiers for jobs and products. Errors that still occur are returned to their source with a reason, so patterns become visible and fixable. Finance reviews exceptions rather than repairing everything.

For the engineering services firm, that might mean job codes selected from the active job list rather than typed, purchase orders required before supplier invoices can be approved, and a monthly report of corrections by branch shared with branch managers.

Intervention

1. Log the corrections. For one month-end, record every correction finance makes, by type and by source department or system.
2. Rank by volume and effort. A small number of error types usually account for most of the time.
3. Move validation upstream. For each top error type, add a rule, default, required field or restricted list at the point of capture.
4. Assign master data ownership. Name an owner for customers, suppliers, products and cost centres, and control who can create them. The Insight How to Decide Which System Owns Which Data helps here.
5. Close the feedback loop. Report errors to the departments that create them, regularly and without blame.
6. Measure the close. Track close duration and correction volume month by month.

Diagnostic questions

• How many working days does your month-end take, and how many of them are spent correcting data?
• Which three types of error does finance fix most often?
• Can anyone in the business create a new customer or supplier record?
• Do the people who make capture errors ever find out?
• Does your finance team keep correction spreadsheets or lookup tables outside the accounting system?

Bottom line

When finance teams become data clean-up teams, the business is paying skilled people to repair errors that its systems allowed in. Moving validation to the point of capture, giving master data a clear owner and returning errors to their source shortens the close, improves trust in the numbers and frees finance to do the work only finance can do.

Sources and further reading

• Data cleansing - Wikipedia (last verified 10 October 2026)
• What Is Data Quality? - IBM (last verified 10 October 2026)
• Section 16 Quality of information - POPIA (last verified 10 October 2026)

Related Mellorca reading

• Why Does Sending an Invoice Require So Much Work?
• When Cash Collection Depends on Someone Remembering to Follow Up
• Why Returns Become a Cross-Department Investigation