How do we know the reports match reality? That question sits behind every board meeting, every investor conversation, and every audit. Reconciliation is the process that answers it — not with assertion, but with evidence.
Yet in many mid-market finance teams, reconciliation has been quietly replaced by something that looks similar but functions differently: explanation. The numbers do not match, so someone writes a commentary about why. The difference is rationalised, not resolved. The report goes out, and the underlying discrepancy remains. This article examines why genuine reconciliation matters, what it consists of, and how to build it into a repeatable process that does not depend on any single person’s knowledge.
What reconciliation is — and what it is not
Reconciliation is the process of comparing two or more data sets to verify they agree, and then identifying and resolving any discrepancies. It is a verification activity: does the report match the source? Do the systems agree? Does the movement between periods make sense?
The word “resolving” is doing important work in that definition. Reconciliation is not complete when a difference is found. It is complete when the difference is either corrected or explained with a documented, evidenced justification that accounts for the precise amount. An unexplained or hand-waved difference is an open item, not a reconciled item.
There is a useful distinction between three types of activity that often get conflated:
| Activity | What happens | Outcome |
|---|---|---|
| Reconciliation | Compare data sets, investigate differences, resolve or document with evidence | Verified, trustworthy numbers |
| Explanation | Note that numbers differ and write a narrative about possible reasons | A story, not verification |
| Acceptance | Observe that numbers are “close enough” and proceed | Unverified numbers treated as verified |
Many finance teams operate somewhere between explanation and acceptance while believing they are reconciling. The distinction matters because explanation and acceptance allow errors to persist and compound across periods.
Why reconciliation matters — the business case
The cost of unreconciled data
Finance teams spend 5–10 working days per month reconciling data. Much of this time is rework — chasing differences that originated in poor upstream processes, correcting errors that should have been caught at the point of entry, and reconstructing data flows that were never documented.
The Hackett Group benchmarking shows that top-quartile organisations with documented reconciliation processes spend 30% less time on reconciliation activities than their peers. The difference is not that top performers skip reconciliation — it is that their upstream processes produce fewer discrepancies to reconcile.
Speed of close
FloQast (2025) reports that the average mid-market close takes 10–15 working days. Best practice is five or fewer. The primary differentiator between these two groups is not accounting complexity or team size — it is reconciliation efficiency.
Deloitte’s “Fast Close” research confirms the downstream impact: companies closing in five days see audit fees approximately 40% lower than slower closers. Faster close requires that reconciliations are systematic, documented, and largely completed as part of the ongoing process rather than concentrated in the close window.
Trust and decision quality
When leadership receives reports built on unreconciled data, two things happen. First, errors affect decisions — a margin figure that includes an unresolved inter-company difference leads to wrong conclusions about profitability. Second, when those errors surface later, leadership trust in the numbers erodes. Once trust is lost, every subsequent report is questioned, and the finance team spends more time defending numbers than analysing them.
Audit evidence
Reconciliations are primary audit evidence. External auditors examine reconciliations to verify that reported balances are supported by underlying records. An unreconciled balance is an audit finding. A reconciliation with unexplained differences is a qualified finding. A well-documented reconciliation with all items resolved is evidence that the internal controls are functioning.
Types of reconciliation
Source-to-report reconciliation
Does the report match the source data? This is the most fundamental form of reconciliation. The general ledger says revenue is £4.2M. The management report says £4.2M. If they differ, something happened in the extraction, transformation, or presentation — and that something needs to be identified and resolved.
Source-to-report reconciliation catches errors in data extraction, calculation logic, mapping tables, and manual adjustments. It should be performed every reporting cycle for all material line items.
Inter-system reconciliation
Do different systems agree? The ERP says accounts receivable is £1.8M. The CRM says outstanding invoices total £1.9M. The bank says receipts this month were £0.6M. Inter-system reconciliation verifies that these related numbers are consistent, or explains precisely why they differ (e.g., timing of cash receipts, credit notes not yet posted).
This is where the “five versions of the truth” problem lives. When finance, sales, and operations each extract data from different systems without reconciling across them, the organisation produces multiple sets of numbers that all claim to be correct.
Period-to-period reconciliation
Does the movement between periods make sense? Revenue was £4.0M last month and £4.2M this month. A £200K increase is plausible given seasonal patterns. A £2M increase warrants investigation. Period-to-period reconciliation uses reasonableness checks to identify anomalies that may indicate errors.
This form of reconciliation is particularly effective at catching offsetting errors that balance reconciliation would miss. Two errors of £500K in opposite directions produce a correct total but an incorrect composition — period-to-period analysis surfaces the unusual movements.
Balance reconciliation
Do components sum correctly? The three divisional revenue figures should sum to group revenue. Cost of goods sold plus gross profit should equal revenue. Balance reconciliation verifies internal consistency — that the mathematics of the report hold.
This is the simplest form of reconciliation and the one most often automated. It is necessary but not sufficient — a report can be internally consistent while being consistently wrong if the source data is incorrect.
Building a reconciliation process
Frequency matched to risk
Not every reconciliation needs to happen at the same frequency. A practical approach matches frequency to volume and risk:
| Reconciliation type | Suggested frequency | Rationale |
|---|---|---|
| High-volume transaction matching (e.g., bank) | Daily | Errors compound quickly; daily matching keeps the backlog manageable |
| Inter-system checks (e.g., ERP to CRM) | Weekly | Catches drift before it becomes material |
| Source-to-report for management accounts | Monthly (at close) | Ensures every reported number is verified |
| Period-to-period reasonableness | Monthly | Identifies anomalies while context is fresh |
| Full balance sheet reconciliation | Monthly | Audit requirement; prevents balance build-up |
Materiality thresholds
Not every difference warrants full investigation. Materiality thresholds define the point at which a difference must be investigated versus accepted and documented. A £0.12 rounding difference in a £4M revenue line is immaterial. A £12,000 difference is not.
Thresholds should be:
- Defined in advance (not decided ad hoc when a difference is found)
- Approved by the finance director or controller
- Documented and applied consistently
- Reviewed periodically as the business grows
Exception handling and escalation
When a reconciliation produces a difference above the materiality threshold, there must be a defined process: who investigates, what the deadline for resolution is, and when the issue escalates. Without escalation rules, exceptions accumulate in a spreadsheet that nobody reviews — and eventually become the items that auditors find.
A simple escalation structure:
| Difference size | Action | Escalation |
|---|---|---|
| Below materiality | Document and accept | None required |
| Above materiality, below 1% of line item | Investigate within 5 working days | Finance manager |
| Above 1% of line item | Investigate within 2 working days | Finance director |
| Above 5% of line item or recurring | Immediate investigation | CFO |
Documentation
Every reconciliation should produce a record that answers:
- What was compared?
- What was the result?
- Were there differences? If so, what were they?
- How were differences resolved?
- Who performed the reconciliation and when?
This documentation serves two purposes. First, it provides audit trail evidence. Second, it enables someone other than the person who performed the reconciliation to understand and verify the result. If only one person can interpret a reconciliation workbook, the process has a single point of failure.
Common pitfalls
Reconciling totals only
Comparing the total in the report to the total in the source and concluding “they match” misses offsetting errors. Revenue could be overstated by £100K in one category and understated by £100K in another — the total is correct, but the composition is wrong. Reconciliation must operate at the level of detail appropriate to the risk, not just at the summary level.
Explaining away differences
Writing a narrative about why numbers differ is not reconciling them. “Revenue is lower because of seasonal patterns” may be true, but it is a hypothesis, not a verification. Reconciliation investigates the actual transactions that produced the difference — not the plausible story that might explain it.
Person-dependent reconciliation
When only one person knows how to reconcile a particular balance — which spreadsheet to use, which adjustments to make, which exceptions are “normal” — the organisation has a risk, not a process. If that person is unavailable during the close, the reconciliation either does not happen or is performed by someone guessing at the method.
Every reconciliation should be documented well enough that a competent finance professional who has never performed it before can follow the steps and reach the same result.
No escalation for unresolved items
Without a defined threshold for when a difference must be escalated, exceptions get buried. The reconciliation spreadsheet shows a £45K unresolved difference that has been carried forward for three months. Nobody escalated it because there was no rule requiring escalation. The auditor finds it and raises a management letter point.
Manual reconciliation that does not scale
A manual reconciliation that works for 200 transactions per month breaks down at 2,000. As the business grows, reconciliation processes must scale — through automation of routine matching, exception-based workflows, and appropriate use of technology for high-volume activities.
Technology perspective
Technology does not replace the need for reconciliation judgement, but it eliminates the manual effort in routine matching. Automated reconciliation can match thousands of transactions per minute against defined rules, leaving the finance team to focus on the exceptions that require investigation.
Relevant capabilities include:
- Automated matching — rule-based comparison of transactions across data sets, with configurable tolerance and matching criteria
- Exception management — workflow-based routing of unmatched items to the appropriate person, with ageing tracking and escalation
- Auto-generated documentation — reconciliation logs produced as a by-product of the matching process, eliminating manual documentation effort
- Status visibility — real-time view of which reconciliations are complete, which have open items, and where the close process stands
The documentation element is particularly important for reducing person-dependency. When the reconciliation process auto-generates its own records, those records survive personnel changes — unlike the knowledge that currently lives in one person’s head.
Frequently asked questions
What is the difference between reconciliation and data validation ? Data validation checks whether individual data points meet defined quality criteria — is the value within range, is the field populated, is the format correct. Reconciliation compares data sets against each other to verify consistency. Validation catches individual errors; reconciliation catches systemic discrepancies between sources.
How do we reconcile when the source systems use different structures? This is common — the ERP uses one chart of accounts, the CRM uses different categories, the bank statement has its own format. The reconciliation process needs a mapping that translates between structures. Document the mapping, version-control it, and treat changes to the mapping as controlled changes that require approval.
What materiality threshold should we use? There is no universal answer. Thresholds depend on the size of the balance, the nature of the account, and the risk tolerance of the organisation. A starting point: investigate differences above 1% of the line item value, or above a fixed amount that is material to the business (e.g., £5,000). Refine based on experience.
How do we stop carrying forward unresolved items? Ageing rules. Define a maximum age for an open reconciliation item — typically 30 to 60 days. Items older than the threshold must be either resolved or escalated to the finance director with a documented reason for the delay. Make the ageing report a standing item in the monthly close review.
Can reconciliation be fully automated? Routine matching can be automated — bank reconciliation, invoice matching, inter-company elimination. But exception investigation, judgement about materiality, and resolution of complex differences require human analysis. The goal is to automate the 80% that is routine so the finance team can focus on the 20% that requires thought.
Related Reading
- Data Governance in Financial Reporting — governance as the prerequisite for reconciliation that works
- Audit Trail and Traceability in Reporting — the documentation that reconciliation produces and audit requires
- Financial Data Quality Warning Signs — symptoms that indicate reconciliation is failing
- Month-End Close Best Practices — the close process where reconciliation lives
- Internal Controls for Mid-Market Companies — controls that depend on effective reconciliation
- Single Source of Truth in Finance — what reconciliation is working toward
- Variance Analysis Guide — analysis that depends on reconciled numbers
- Glossary: Reconciliation | Data Validation | Audit Trail
Sources
- FloQast — 2025 Controller’s Guidebook — average mid-market close 10–15 days; best practice 5 or fewer
- Deloitte — “Fast Close” Research — companies closing in 5 days see ~40% lower audit fees
- The Hackett Group — top-quartile organisations with documented processes spend 30% less time on reconciliation
- Industry benchmarks — finance teams spend 5–10 working days per month on reconciliation activities
Martin Duben is a finance and reporting specialist at Onetribe. He works with mid-market companies across Central Europe to build reporting infrastructure that produces trustworthy, reconciled numbers — from source data through to board pack.