Key Takeaways

• Set reconciliation policy once at group level and let each entity apply it. Intercompany is where that split fails first: every break belongs to two entities, so both sides need to see the same status.

• A reconciliation is only as current as the data behind it, so connect your enterprise resource planning (ERP) system and subledgers directly. On bank feeds, plan for MT940 and camt.053 files in the same close for as long as any of your banks still sends MT940.

• Tier every account by volume and judgement before you automate anything. The tier decides the matching method, the review depth and the frequency.

• Accuracy at scale comes from tolerance thresholds, recorded match logic, period sampling and automatic reopening. Give every exception a named owner and an ageing clock, or it becomes an audit question.

Scale breaks the coordination around account reconciliations long before it breaks the matching. Sooner or later, every growing group reaches the same fork in the road and has to choose a path. You hire another reconciler for each new entity, or you redesign your reconciliation workflows across entities so the existing team covers more ground.

Hiring works for a year or two, because while the entity count is small, extra people can absorb the extra accounts each new entity brings. Then the spreadsheet tracker and the email chasing become the bottleneck, and the close stops getting shorter however good the matching is.

Most advice starts in the wrong place, by asking which tool to buy. In our experience the operating model matters more: who owns each account in each entity, and how you prove the work was right. This article answers five questions: how to run the workflow across entities, how to automate across subledgers and bank feeds, how to approach high-volume, low-complexity reconciliations, how to protect accuracy, and how to track ownership.

How Do Finance Teams Manage Reconciliation Workflows Across Entities?

Finance teams manage reconciliation workflows across entities with one group policy, one template set, a central status view and a named owner per account per entity. Each entity is configured locally, so the group controller sees one picture of the close instead of a tracker per entity.

Group policy and local execution are separate layers: the group sets frequency, thresholds, evidence standards and ownership rules once, and each entity applies them. Intercompany tests this design first, because every break belongs to two entities. That is why intercompany matching across entity pairs needs status visible to both sides.

What Breaks When Volume Scales

Each new entity brings more bank accounts and subledgers. It also adds intercompany pairs: ten entities trading with each other can create up to 45. Reconciliation effort therefore rises faster than the account count.

Groups that grow by acquisition often end up running several ERP systems, and Deloitte's Oracle alliance overview notes that the question of how to aggregate data from multiple ERP systems arises in nearly every mergers and acquisitions (M&A) transaction. A typical pattern is NetSuite or Microsoft Dynamics 365 at group level, with Visma, Fortnox, Exact or Sage in acquired subsidiaries, each with its own chart of accounts and close calendar.

Close duration grows with group size too. In our experience, most of that growth is coordination across more entities and more owners, not extra matching.

Before you evaluate any tool, count your entity pairs, banks, bank formats and external subledgers. That count sizes the problem better than the number of general ledger accounts.

How Can Finance Teams Automate Account Reconciliations Across Subledgers and Bank Feeds?

Connect each source directly so reconciliations run on live data, not month-end exports, and make sure your matching reads every bank format your entities receive. Most of the effort goes into tying each source's cut-off back to its control account.

The subledgers rarely sit inside the ERP. Card and payment service provider (PSP) settlements arrive from Stripe or Adyen, billing from Chargebee and spend from Coupa or Pleo. Payroll comes from a separate provider again, and each source has its own cut-off to tie back to a control account.

Stale exports are the quietest failure. A preparer exports the ledger on working day two and a late journal posts on day three. The reconciliation now ties to a ledger that no longer exists.

Bank feeds add a format problem. Some banks already deliver camt.053, the ISO 20022 statement format, while others still send MT940, and each bank sets its own switch-off date. DZ BANK's camt.053 migration notice runs both formats in parallel until 30 November 2026. Danske Bank's upcoming changes page says MT and camt statements will continue to be supported. Swift's Cross-Border Payments and Reporting Plus (CBPR+) programme ends MT/camt coexistence for interbank statements in November 2028, according to ING's CBPR+ migration update, but it doesn't change a corporate's own format. Whether you receive both formats, and for how long, depends on each bank's own timetable, so ask each bank what it will send you. In a demo, load an MT940 and a camt.053 file for the same account and check that both match to the same ledger lines.

What's the Best Approach to Automating High-Volume, Low-Complexity Reconciliations?

Tier accounts on two axes before automating: transaction volume, and the judgement needed to conclude on the balance. The two axes give four quadrants, and each quadrant needs its own matching method and review regime.

Our enterprise guide to reconciliation automation covers rollout sequencing in more depth. The matrix below is the version we use in implementations, with accounts receivable (AR) and accounts payable (AP) control accounts in the high-volume tier.

Source: Stacks implementation practice, 2026.

Automating the High-Volume Tier

The high-volume tier is bank and PSP clearing (card settlements included), plus AR and AP control accounts tied out to their subledgers. Deterministic rules handle exact matches on amount and reference. Learned matching handles what rules miss: one-to-many splits, batched settlements, partial payments and FX differences.

Set the target at 95%+ auto-match on this tier, and expect to reach it through learned matching rather than an ever-longer rule list. Stacks' comparison of legacy tools puts rules-based approaches at 70–75% match rates, with rules that need constant tuning.

Start with the cleanest, highest-volume accounts and agree the exception policy before go-live.

In a pilot, run your current matching and the new engine side by side on your busiest bank accounts for one close. The items where they disagree tell you more than any vendor's headline match rate.

Keeping Judgement Where It Belongs

Accruals, provisions, intercompany disputes and FX revaluation need a conclusion that software should not reach alone. Automate the gathering instead. Pull the balance, attach the supporting schedule, roll forward last month's workings and flag movements above threshold.

The preparer then concludes, and the reviewer challenges the conclusion. A tool that proposes a provision balance with no reviewer in the loop is a plausible-looking wrong number waiting to reach the accounts.

Frequency should follow risk. We recommend reconciling high-risk accounts monthly and low-risk accounts quarterly, which frees reviewer time for the quadrant where errors are expensive. Record the risk rating per account, so your auditor can see why a quarterly account had no reconciliation in month two.

How Do You Keep Reconciliation Accuracy High as Automation Increases?

Accuracy at scale comes from workflow design: thresholds per tier, a recorded basis for every match, period sampling and automatic reopening. The matching engine, whether rules or artificial intelligence (AI), is one input to that design.

Adoption is already high. Gartner's 2025 finance AI survey of 183 chief financial officers (CFOs) and senior finance leaders found that 59% use AI in their function. Yet Gartner's CFO AI roadmap research, drawing on the same survey, found only 7% report high or very high impact. In our experience, that gap opens when AI is bolted onto an unchanged workflow, which is exactly the risk with reconciliation matching.

Validation, Thresholds and Spot-Checks

Set tolerance and materiality thresholds per tier, and publish them in the group policy so no preparer sets their own. A bank clearing account might auto-certify small fixed differences, while a provision account gets no auto-certification at all.

Every auto-match should carry its basis: the rule or model that matched it, and any tolerance applied. Each period, sample auto-cleared items in each tier and re-perform them, so you hold evidence that the engine is still right.

If a balance changes after sign-off, the reconciliation should reopen automatically and alert the right people. Expect to tune rules over the first two or three cycles, and log each change so the reviewer can see why the match rate moved.

The failure to design against is a reconciliation that balances for the wrong reason. Two offsetting errors net to zero, or a plug sits in a suspense line, and the sign-off looks clean. It is the same failure mode behind many spreadsheet failures in finance: nothing crashes, so nobody looks.

Evidence a Board Can Rely On

The Financial Reporting Council (FRC) has brought the evidence question into the boardroom for UK Corporate Governance Code companies. Under the FRC Provision 29 Mythbuster, the revised Provision 29 applies to accounting periods beginning on or after 1 January 2026. Boards declare the effectiveness of material controls as at the balance sheet date, and the FRC says reporting starts from 2027.

The Code names no specific control, and the FRC does not say which controls are material. In our view, key balance sheet reconciliations are a likely candidate wherever a board treats financial reporting controls as material.

A declaration needs evidence that the control operated, captured when the work happened. Sign-offs rebuilt from email threads in January are weak ground for a board to stand on.

Test your position before year-end. Pick a reconciliation from three months ago and check whether an outsider could see who prepared and reviewed it, and what changed afterwards.

How Should Ownership and Progress Tracking Work Across a Finance Team?

Every account in every entity needs a named preparer and a named reviewer, with segregation of duties enforced by the system. Ownership set at that level removes the question we hear most often in a slipping close: whose is this?

Exceptions need owners as well, so each unmatched item becomes a work item with an escalation service level agreement (SLA) tied to its age. Old breaks then show up in the queue, not in an email thread three closes later.

Tracking Progress Without a Spreadsheet Tracker

Progress should come from the work itself. A status view tied to a live close calendar tells you where the close stands by entity and by owner. We recommend tracking four states: not started, in progress, blocked and overdue.

The same view lets you rebalance workload mid-close. If one reviewer holds most of the open sign-offs on working day three, reassign them that day, while there is still time to clear them.

Track four numbers every close:

• Auto-match rate by tier and entity

• Exception count and ageing by owner

• Reconciliations signed off by your target working day

• Reconciliations reopened after sign-off, with the reason

Measure all four before a pilot and again after two closes. If the sign-off curve has not moved earlier, the tool changed the matching without changing the workflow.

Why Standard Advice Fails

The common playbook is to buy a matching tool and keep the spreadsheet tracker. It looks sensible, because matching is the most visible manual effort and the tracker already exists.

In our experience, matching improves quickly. Yet exceptions still get chased over email, and the reviewer still assembles evidence by hand at quarter-end. The team now matches most transactions automatically and still closes on the same day it always did.

There is a fair case for the playbook, since a group with two entities and one bank can run a tracker well. The case breaks once ownership spans more people than one controller can hold in their head. In our experience, that happens sooner than most groups expect.

When you evaluate vendors, ask for these in the demo:

• Show an exception moving from detection to reviewer sign-off.

• Change a ledger balance after sign-off and watch what happens to the reconciliation.

• Open an auto-cleared item and display the basis for the match.

• Pull up one intercompany pair and compare its status in both entities.

If the demo stops at the match, so does the tool.

Why Choose Stacks for Reconciliations at Scale

Stacks connects to your ERP and syncs ledger data live, so every reconciliation starts from current balances. On that base, our reconciliation automation for enterprise matches above 95% of transactions out of the box, and the match rate improves with every period the engine sees.

Exceptions become tracked work items with an owner and a status, and every reconciliation has a defined preparer and reviewer, which is the design behind centralised reconciliations in Stacks. The audit trail builds as the work happens, so the evidence an auditor or board needs already exists when they ask.

Juni's finance team has seen what that does to the close. Ruben A., Head of Finance at Juni, puts it plainly: "Since using Stacks, we've reduced the time to financial close by three and a half days, which is material in our case." [VERIFY: period or date of Juni's 3.5-day reduction, and whether a Juni customer story URL exists]

Andy M., Head of Finance at Cleo, sees the effect further downstream: "With Stacks, we're saving a noticeable amount of time on the financial close, and as a result we can produce the management accounts faster each month."

FAQs for Account Reconciliations at Scale

What Auto-Match Rate Should We Expect from Reconciliation Automation?

Aim for 95% or more on high-volume bank and PSP clearing accounts once the first two or three closes have tuned the rules. Stacks' legacy tool comparison puts rules-based approaches at 70–75%.

Which Reconciliations Should We Automate First?

Start with your cleanest, highest-volume accounts, usually bank and PSP clearing. They give the fastest gain and build confidence before you move to subledger tie-outs.

How Do Continuous and Periodic Reconciliation Differ?

Periodic reconciliation checks a balance at month-end or quarter-end. Continuous reconciliation matches transactions as data arrives, so month-end becomes a review of exceptions.

How Often Should Low-Risk Accounts Be Reconciled?

Quarterly is reasonable for low-risk accounts, with monthly reconciliation for high-risk ones. Record the risk rating so your auditor can see the basis for each frequency.

What Signs Show a Team Has Outgrown Manual Reconciliation?

Exceptions that age across closes without an owner, and sign-offs that bunch on the final day, are the clearest signs. If adding an entity means adding a person, the workflow has stopped scaling.

See How Stacks Maps to Your Close Process

Book a 30-minute walkthrough. We'll discuss your close process, show you which tasks run automatically, and outline the next step from there.

GET STARTED