Data Quality and the Finance Function

Why data quality matters in finance: master data, causes of poor data, ownership, controls that work, metrics, AI and system migration.

  • Founded by Adrian Lawrence FCA, ICAEW
  • Permanent, interim and fractional finance hires
  • Offices in London and the West Midlands
Finance team meeting at a table

Why does data quality matter in finance? Every report, reconciliation and forecast is built on master data (suppliers, customers and the chart of accounts) and on the transactions coded against it. Poor data shows up as duplicate payments, failed reconciliations, management accounts that need reworking each month and system migrations that overrun. Good finance data quality comes from controls at the point of entry, regular reconciliations and clear ownership of each data set.

Every number the finance function produces rests on the quality of the data underneath it. Where that data is poor, the reporting is unreliable however good the people and however modern the systems; where it is good, the function can close quickly, trust its numbers and spend its time on analysis rather than correction. Data quality is not a peripheral IT matter. It is one of the main determinants of how well a finance function performs, and it has become more important as businesses automate processes and introduce AI.

This guide is written for Financial Controllers, finance managers and the accountants who do the work. It covers what good data looks like, where poor data comes from, the master data that matters most to finance, who should own it, the controls and metrics that keep it clean, its effect on automation and AI, how to protect it through a systems migration, and which roles to hire when the problem is bigger than the current team can absorb.

Why data quality matters so much to finance

Finance turns data into information, and the information cannot be more reliable than the data. Poor data quality undermines the function in four ways. The reporting is wrong or late because errors must be found and corrected first. The analysis is weaker because it works from figures nobody fully trusts. Decisions taken on those figures are worse. And the time spent correcting and working around bad data is time not spent on the work the business actually values.

The cost is often hidden. It shows up as a month-end that never quite gets faster, a suspense account that never clears, reconciliations that need “adjusting”, customer statements that customers dispute, and a board that asks for numbers to be re-checked. Each symptom looks like a process problem; underneath, many are data problems. Our guide to optimising the month-end close covers the process side, but a close built on poor master data stays slow however well the timetable is designed.

There is a regulatory dimension too. Under Making Tax Digital for VAT, records must be kept digitally and data moving between software must flow through digital links, so a spreadsheet re-keyed by hand is a compliance issue as well as a quality one. The government has also announced that e-invoicing will become mandatory for VAT invoices from April 2029, following its consultation on electronic invoicing; structured invoices exchanged between systems will expose inconsistent supplier and customer records quickly. And where finance data includes personal data, such as individual customers or employees, the UK GDPR’s accuracy principle requires it to be kept accurate and up to date.

What good and poor data quality look like

Good data has five qualities:

  • Accurate: it correctly represents what it describes, such as the right bank details for a supplier or the right VAT treatment on an account.
  • Complete: required fields are filled, such as cost centre, project code or customer company number.
  • Consistent: the same thing is recorded the same way everywhere, so a customer has one record across the CRM, billing system and ledger, not three slightly different ones.
  • Timely: it is available when needed and kept up to date, so accruals, credit notes and supplier changes land in the right period.
  • Valid: it conforms to the rules and formats it should, such as a nominal code that exists, a date in the period, or a VAT number in the correct format.

Poor data is the absence of any of these, and it shows up as duplicate suppliers, conflicting customer records, missing codes, transactions posted to the wrong period or entity, and balances that cannot be explained. The value of defining good data this way is practical: you cannot improve what you cannot describe, and each dimension can be measured, as the metrics section below shows.

Master data: suppliers, customers and the chart of accounts

Transactional errors are usually one-offs. Master data errors repeat on every transaction that uses the record, which is why master data is where effort pays back fastest.

Supplier master data

The supplier file carries the highest fraud and error risk in most businesses. Duplicate suppliers allow duplicate payments; unverified bank detail changes are a standard route for invoice-redirection fraud; missing VAT numbers or incorrect VAT status cause input tax errors. The essentials are a single controlled process for creating and amending suppliers, independent call-back verification of any bank detail change using contact details already held rather than those on the request, segregation between the person who amends supplier records and the person who approves payments, and a periodic review that removes dormant and duplicate records. Our guide to internal controls for growing businesses sets this in the wider control framework.

Customer master data

Customer records drive invoicing, credit control, revenue reporting and, increasingly, customer metrics. Inconsistent records mean the same customer appears under several names, credit limits are set against the wrong entity, and group exposure is understated. In subscription businesses the problem extends into recurring revenue and retention reporting, because cohort metrics depend on customers being grouped consistently; our guide to SaaS metrics for management accountants shows how quickly NRR goes wrong when they are not.

The chart of accounts and reporting dimensions

The chart of accounts is master data too, and one of the most neglected. Codes accumulate over years, near-duplicate accounts appear (“Sundry”, “Miscellaneous”, “Other costs”), and departments, cost centres and projects are used inconsistently. The result is management accounts that need manual reclassification every month. A good chart of accounts is short enough to be used correctly, has clear descriptions of what each code is for, maps cleanly to the statutory accounts and the management pack, and is changed only through a controlled request. Reviewing it is usually the first step before any reporting or systems project.

The causes of poor data quality

Poor data has identifiable causes, and addressing them at source is far cheaper than correcting the symptoms every month:

  • Data entry without validation: free-text fields where drop-downs should be, no mandatory fields, no format checks.
  • Too many systems, poorly integrated: CRM, billing, payroll, expenses and the ledger each holding their own version of the truth, with manual uploads between them.
  • Spreadsheets as systems: key data maintained outside controlled systems and re-keyed.
  • No clear ownership: everyone uses the data, no one is accountable for it being right.
  • Errors carried forward: bad data migrated from one system to the next because cleansing was cut from the project timetable.
  • Accumulation over time: dormant records, legacy codes and unreconciled balances nobody has had time to clear.

The encouraging point is that almost all of these are within the finance function’s influence. They come down to controls, process, ownership and discipline rather than to anything exotic.

Data ownership: who is responsible for what

Data quality fails most often because ownership is assumed rather than assigned. A workable model separates three roles. The data owner is accountable for a data set being fit for purpose and approves the rules (for example, the Financial Controller owns the chart of accounts and the supplier master). The data steward maintains it day to day under those rules (often a purchase ledger lead or systems accountant). Data users enter and rely on it and are responsible for following the rules.

Finance rarely owns all the data it depends on. Customer data may be owned by sales operations, employee data by HR, product and pricing data by the commercial team. Where finance depends on another team’s data, agree the definitions, the fields finance needs, and how errors are reported and fixed, and record that agreement. The Financial Controller is usually the right person to convene this, which is one reason data capability features in our guide to designing financial controls.

Improving and maintaining data quality: controls that work

Improvement normally needs a one-off clean-up of accumulated problems, followed by controls that stop them returning. The controls fall into three groups.

Preventive controls: validation at the point of entry

  • Mandatory fields and drop-down lists for coding, VAT treatment and dimensions.
  • Duplicate checks on supplier and customer creation (name, VAT number, company number, bank details).
  • Locked periods, so postings cannot land in a closed month without approval.
  • Approval workflows for master data changes, with segregation from payment authority.
  • Purchase order matching, so invoices arrive pre-coded against an approved commitment.

Detective controls: reconciliations and reviews

  • Bank, control account and intercompany reconciliations completed monthly, reviewed, and with reconciling items aged and cleared.
  • Sub-ledger to general ledger reconciliations for receivables, payables, fixed assets and payroll.
  • Exception reports: duplicate invoice numbers, payments to recently amended bank details, postings to suspense or rarely used codes.
  • Periodic review of master data for dormant, duplicate and incomplete records.

Corrective discipline

When an error is found, correct it at source rather than with a top-side journal, record why it happened, and fix the cause if it recurs. Journals that “correct” the same issue every month are a sign that the control is missing upstream.

Data quality metrics

Measuring data quality turns it from an opinion into a managed number. A short scorecard, reviewed monthly by the Financial Controller, is usually enough:

Metric What it tells you
Duplicate or near-duplicate supplier and customer records Master data control and duplicate payment risk
Percentage of records with mandatory fields complete Completeness of master data
Supplier bank detail changes verified by call-back Fraud control operating
Balance and age of suspense and unallocated items Coding accuracy and timeliness
Number and value of unreconciled items over 30 days Whether reconciliations are effective
Bank auto-match rate Consistency of references and customer data
Manual and correcting journals per month How much is fixed after the event rather than at source
Invoices received without a purchase order Upstream process compliance
Post-close adjustments Reliability of the first-cut numbers

Pick five or six that match your known problems, set a baseline, and track the trend. The direction matters more than the absolute number.

Need someone to fix the data underneath the numbers? — 0204 553 8893

We provide Financial Controllers, financial and management accountants and systems-minded finance professionals, permanent, interim and fractional. We respond the same working day; interim candidates typically within days.

Tell us about your hire →Register as a candidate →

Data quality in the modern finance function: automation and AI

Automation applies the same processing to every record, so it processes good data efficiently and bad data equally efficiently. A rules-based bank matching tool fed with inconsistent customer references produces a long exceptions list; an automated accrual built on a poorly maintained chart of accounts posts to the wrong codes every month without anyone noticing.

AI raises the stakes again. Tools that suggest coding, match transactions or draft commentary learn from and operate on the data they are given, and their errors can be harder to spot because the output looks plausible. Our assessment of what is real and what is hype in finance AI makes the same point from the technology side: the businesses getting value are those with clean, consistent data and documented processes. For reconciliations specifically, our guide to AI for reconciliations explains why match rates depend on reference and master data quality more than on the tool.

The ICAEW’s technology resources for accountants return to data governance repeatedly for the same reason. Three practical implications follow. Clean the master data before automating the process that uses it. Keep a human review step on outputs where errors would be material; our guide to human-in-the-loop controls covers how. And control what data AI tools can reach; connecting AI to finance data safely deals with access and security. The wider AI in finance hub links the rest of our guidance, including how the management accountant’s role is changing with automation.

Data quality and systems migration

A new ERP or finance system is the moment data quality is either fixed or locked in for another decade-long system life. Migrations routinely run late because cleansing was under-scoped, and the most common post-go-live complaint is that the new system reports the old problems faster.

The practical sequence is to cleanse before you migrate, not after: rationalise the chart of accounts, merge duplicate suppliers and customers, close dormant records, and clear old reconciling items. Decide how much history to migrate (often opening balances plus a limited period of detail, with the old system archived for reference). Reconcile migrated balances to the legacy ledger at every trial load, and have finance, not the implementation partner, sign off the reconciliations. Our guide to migrating finance systems without disruption covers the project plan in full, and Financial Controllers for ERP and system change covers the leadership role. Where the permanent team cannot absorb the extra work, an interim Financial Controller for a systems implementation is a common answer.

Building a culture that values data quality

Controls set the floor; culture decides whether data stays clean between reviews. The people who enter and use data maintain its quality, and where they understand why it matters and own the data they handle, it stays clean. Where data quality is seen as someone else’s job, it degrades despite the controls.

Building that culture is mostly about visibility and consistency: explaining to purchasing, sales and operations teams how their data feeds the numbers, sharing the data quality scorecard, recognising teams who improve, and having finance leaders treat data errors as a cause to fix rather than a journal to post. A function takes its cue from its leadership; a Financial Controller who visibly prioritises data quality builds a team that does too. Our finance transformation roadmap places data quality as an early stage for the same reason.

From the team

What we see when data quality is the real problem

The brief says close, the problem is data

Businesses often ask us for a Financial Controller because the month-end takes too long. When the new FC arrives, the cause is frequently the data: duplicate suppliers, an overgrown chart of accounts and cost centres nobody owns.

System changes bring it to the surface

A new finance system is when poor data costs most. Interim help to clean and map the data before migration is usually cheaper than fixing it after go-live.

The Accountancy Capital team, with Adrian Lawrence FCA overseeing each engagement. View Adrian’s ICAEW profile

Who to hire to fix data quality

The right hire depends on whether the problem is ownership, capacity or technical capability.

Systems accountant

A systems accountant (sometimes titled finance systems accountant or finance systems analyst) sits between finance and IT. They maintain the ledger configuration and master data, build reports, manage integrations and lead the finance side of system changes. This is the role to consider when the finance system itself is the source of problems, or when an ERP project is approaching. Salaries vary widely with the platform and the scale of the system, so we benchmark each brief individually rather than quote a range.

Financial accountant

A financial accountant owns the balance sheet, reconciliations and control accounts, and is often the natural steward of supplier data and the chart of accounts. House figures in London are £52k–£60k for newly qualified, £60k–£75k with two to five years’ experience, and £72k–£95k at senior or group level, with interim day rates of £275–£400. See the financial accountant salary guide for the detail, and financial accountant recruitment for how we work.

Management accountant

A management accountant cares most about reporting dimensions, cost centres and the chart of accounts, because those determine whether the management accounts need reworking each month. London salaries are £50k–£85k (mid-level £58k–£73k), with interim day rates of £250–£375 in London and £225–£325 elsewhere. The management accountant salary guide has the full benchmarks.

Financial Controller

Where the root cause is that nobody owns data, the answer is usually ownership at controller level: a Financial Controller who sets the rules, convenes the other data owners and runs the scorecard. Our Financial Controller recruitment page covers that hire.

When interviewing for any of these roles, ask candidates to describe a data problem they found, how they traced it to its cause, and what control they put in place. The answer separates people who clear symptoms from people who fix sources.

Frequently asked questions

What is master data in finance?

Master data is the reference information that transactions use: supplier and customer records, the chart of accounts, cost centres, projects, products and tax codes. Errors in master data repeat on every transaction that uses the record, which is why it deserves the tightest controls.

Who should own data quality in the finance function?

Accountability usually sits with the Financial Controller, with named owners for each data set and stewards who maintain it. Where finance relies on data owned by other teams, the definitions and correction process should be agreed in writing.

How do we measure data quality?

Use a small scorecard: duplicates, completeness of mandatory fields, suspense balances, aged reconciling items, manual journals and post-close adjustments. Track the trend monthly.

Should we clean data before or after a system migration?

Before. Migrating poor data moves the problem into the new system, and cleansing is harder once the legacy context has gone. Budget time for it in the project plan.

Do we need clean data before using AI in finance?

For any tool that codes, matches or reports on your data, yes. AI and automation amplify the quality of what they are given; clean master data and documented processes come first.

Adrian Lawrence FCA — Founder, Accountancy Capital

Fellow of the ICAEW | Chartered Accountant holding an ICAEW practising certificate | Part of FD Capital Group

Adrian founded FD Capital in 2018 and later Accountancy Capital, which provides qualified finance professionals below director level. A former listed-company Finance Director with a BSc from Queen Mary College, University of London, he holds an ICAEW practising certificate. Accountancy Capital provides the controllers, accountants and systems-minded finance professionals who keep the data behind the numbers reliable.

View Adrian’s ICAEW profile

Hiring for a cleaner finance function — 0204 553 8893

Financial Controllers, financial and management accountants and systems-minded finance professionals, permanent, interim and fractional. Email recruitment@accountancycapital.co.uk or call our London office.

Tell us about your hire →Register as a candidate →

Start a search

Tell us about your finance hire

Share a few details about the role and your timetable. A consultant will review your brief and come back to you the same working day.

  • Same-day response from a consultant who works on finance searches every day
  • Confidential from the first conversation, with no obligation to proceed
  • Permanent, interim and fractional finance roles across the UK
Please enable JavaScript in your browser to complete this form.

Your details are used only to respond to your enquiry.