All Articles
  • Close & Reconciliation

Balance Sheet Account Reconciliation: Why Your Workpaper and Your Ledger Stop Agreeing

nominal's icon in black on a green and white background

Balance sheet account reconciliation confirms that every figure in the general ledger agrees with the evidence behind it at period end. Gaps are settled by correcting the supporting schedule, posting a journal entry, or doing both, and the outcome is documented for review.

Most controllers describe the same moment in nearly the same words: my Excel says one thing, the GL says another, and I find out on day six. Nothing looked unusual earlier in the month, so nobody opened the file until the calendar was most of the way through.

Balance sheet account reconciliation is the work that catches that mismatch, and the delay rarely comes from the comparison itself. Pulling figures and lining them up against a schedule takes minutes. Deciding what a gap means, and which path settles it, absorbs the rest of the week.

There are only three ways to settle a difference. You correct the supporting file, you post a journal entry, or you do both. Knowing which path applies, and being able to show why, is what turns a flagged number into one that ties.

What Balance Sheet Account Reconciliation Actually Covers

Reconciling the balance sheet means comparing each general ledger figure to the proof behind it, on a set cadence, usually monthly or quarterly. The work confirms the number is right and that anything unexplained gets named, investigated, and cleared inside the month-end close calendar.

The accounts that carry the most risk

Not every line item carries the same risk. Cash is supported by the bank statement, through a formal bank reconciliation. Receivables trace to the aging report. Prepaid expenses depend on the amortization schedule, and fixed assets are checked against the asset register, where this month's figure only makes sense against the one before it.

Each of those brings its own source document, its own pattern of reconciling items, and its own likelihood of error. A team carrying several hundred general ledger lines needs a way to decide which ones deserve attention first and which can run on a lighter cadence.

What supporting evidence means in practice

Nominal interface comparing bank statement transactions against the general ledger with Strong, Probable, and Uncertain confidence scores

Proof is whatever data you would use to defend a figure if an auditor questioned it. For cash, that is the statement. For accruals, it is the schedule plus the contracts or invoices behind the estimate. For deferred revenue, it is the contract file.

The standard is straightforward: another accountant should be able to pick up your working file and trace the number to its origin without asking anyone a question. A note that reads "per management estimate" does not meet it.

The Three Ways a Reconciling Difference Gets Settled

Only three paths close a gap of this kind. Naming the right one early keeps an item from sitting in an open list for a week while two people email each other about it.

1. Correct the supporting file

Sometimes the books are right, and the support has gone stale. An amortization tab nobody updated, a rollforward that never picked up an addition, a mapping sheet built by someone who has since left. The fix happens in the workpaper, the figures agree, and nothing posts.

Your proof still has to show what changed and why, because a corrected version with no explanation attached is the one an auditor asks about six months later.

2. Post a journal entry

Sometimes the support is right and the books are missing something: an accrual nobody recorded, a reversal that never went through, a payment coded where it does not belong. The resolution is a journal entry, prepared against written policy and routed for approval before it posts.

The two sides then agree because the ledger moved, and the approver can see exactly what changed.

3. Do both

The case that consumes the most time is the one where each side is partly wrong. An invoice arrives after the support was built, so the estimate rests on an assumption that no longer holds and nobody revised it. The workpaper needs correcting, the books need a posting, and the two have to agree afterwards.

Spreadsheets handle this path worst, because the correction and the posting are usually made by different people at different points in the month, and nothing connects them except memory.

Where the Workpaper and the Ledger Start to Diverge

Divergence is not random. It comes from a short list of predictable causes, so knowing where to look shortens any investigation.

Timing differences and prepaid entries

Nominal Close Management task showing an automated prepaid expense amortization schedule generated from a vendor bill

Timing explains most of it. A payment goes out in January for a service covering February through April, and the charge is booked on the way out. The amortization schedule lives on a separate spreadsheet updated whenever someone remembers, so the first month of expense never posts and the two sources disagree by exactly that amount.

Missing accruals and unposted transactions

Accruals estimate costs already incurred but not yet invoiced. Without a consistent way to identify and record them at month-end, the books understate expense and overstate income, while the workpaper, built from contract data and prior patterns, reflects where the figure should land.

Reconciling surfaces the gap, though the repair belongs upstream in how estimates get identified and posted each month.

Read More: How to Automate Accrual Reversals, Transaction Matching, and Flux Analysis with AI Agents

Balances that depend on what happened last period

Some general ledger lines cannot be judged on one month alone. A fixed asset register only agrees if the prior closing position carried forward correctly, additions and disposals were both processed, and the depreciation policy was applied per asset class.

Treat each month as independent and those figures drift quietly for a long stretch before anyone notices, which leaves an investigation covering every month since the drift began.

How Accounting Teams Are Automating Account Reconciliation

Nominal Close Management dashboard for December 2025 showing prepaid expense tasks by status — Pending, In Review, Declined, Done — alongside an agent panel covering Payroll, Software Expenses, Prepaid Expenses, Tax Accrual, and other account agents

Figures still have to agree, proof must be complete, and open items require resolution and approval. None of that changes. What has moved is how much of the work a person performs by hand.

Nominal comparison table contrasting Manual Process vs. With Resolution Agents across workflow steps, time per resolution (10–15 min vs. 2–3 min), error risk, audit trail completeness, and logic consistency
  • Rule-based close tools handle the comparison: They pull figures, line them up against sub-ledger or external data, and surface what does not match. When judgment is needed, the tool routes the item to a person and stops there, which leaves the slowest part of the process exactly where it was.
  • A reconciliation agent carries the work further: It compares the figures, identifies the open items, applies the resolution path written policy calls for, corrects the support or drafts the posting, and presents a tied-out result with proof attached. A person approves before anything reaches the books.

What reaches the controller is a reviewable workpaper, a prepared journal entry, and the handful of cases that genuinely need a decision rather than a queue to investigate from scratch.

You might also like: Automated Reconciliation: Why RPA Falls Short at Scale and How APM Agents Finish the Job

Built around your accounting rather than a template

The hardest work stays in spreadsheets because it follows rules somebody wrote for one specific business: how this company treats an in-transit item, which threshold triggers an investigation, what policy says when an invoice lands after the estimate.

Nominal builds the agent against that written procedure. Your team supplies the judgment and the documented process, the finance engineers encode it, and the result runs on your own data. Nothing goes live on guesswork, which is also what makes the output defensible later.

Who stands behind the agent once it runs

Approval routing, traceability, and an audit trail are claimed across the category at this point, so on their own they settle very little. The harder question is who maintains the thing after go-live.

When a process changes, somebody has to update the logic and confirm the numbers still agree. Nominal's finance engineers build each agent, maintain it as the accounting changes, and the approval path and evidence chain arrive with it, rather than as a framework your team assembles and then defends to an auditor.

Recommended read: Human-in-the-Loop AI Governance in Accounting: Thresholds, Escalation, and Audit Trails

A check that surfaces a difference is doing its job. The measure is what follows: how fast the item is routed to the right path, settled, approved, and documented in a form that holds up six months from now.

When balance sheet account reconciliation runs this way, the work arrives finished, the support is reviewable, the posting waits for approval, and proof sits with the result rather than in somebody's inbox. Day six stops being when surprises land, and judgment calls get the hours that used to go to checking.

See what your own month-end looks like when an agent carries the work to sign-off. Book a demo with Nominal.

Meet Agentic Performance Management  Understand the operational shift from tools that assist to agents that execute complete finance workflows autonomously.  Book a Demo
About the writer

Rafael Ataíde is a business development professional with over 12 years of experience, focused on helping finance teams improve efficiency and scale their operations. At Nominal, he works closely with CFOs, controllers, and accounting leaders to identify where manual processes create bottlenecks. With a strong consultative background, he helps teams move beyond spreadsheets by adopting AI agents that execute workflows such as reconciliations, transaction matching, and reporting across multiple systems, reducing manual work, improving accuracy, and enabling finance teams to focus on more strategic activities.

Rafael Ataide, Business Development Representative at Nominal