Prepare a month-end close exception queue
Collect unusual accounting entries for the closing team. Keep each exception connected to its records and a responsible reviewer.
How this works
Collect unusual accounting entries for the closing team. Keep each exception connected to its records and a responsible reviewer. Use this to make the closing team's investigation queue complete and traceable. It should surface missing reconciliations and unusual entries without presenting a generated anomaly label as proof of an accounting error.
Fix the entity, period and data snapshot before running checks. Combine a deterministic close checklist with evidence-backed anomaly explanations. Keep suggested corrections separate from posted entries, and preserve the controller's resolution so reruns do not recreate resolved exceptions.
Before you start
Close checklist
Obtain required reconciliations, materiality rules, owners and deadlines for the entity.
Accounting exports
Collect ledger, subledger and reconciliation snapshots for the same close period.
Resolution register
Retrieve open and resolved exceptions from prior close runs.
Prepare the inputs
- close_scope
- Required
entity, period, currency and cutoff
Prevent records from other entities or periods entering totals.
- ledger_lines
- Required
entry ID, account, date, debit, credit and source
Reconcile export totals before anomaly analysis.
- close_rules
- Required
versioned checks, materiality and owner mappings
Separate mandatory checks from exploratory anomaly suggestions.
- reconciliation_status
- Required
account, completion state and evidence
Missing status means unknown, not completed.
Implementation, step by step
Establish coverage
- Inventory required sources against the close checklist and record extraction times.
- List missing accounts or late subledgers before interpreting apparent differences.
Checkpoint: A coverage table explains what the run can assess.
Run deterministic checks
- Validate debit-credit balance, duplicates, cutoff dates and required reconciliations.
- Use approved materiality and account rules without inventing thresholds.
Checkpoint: Each exception has a reproducible calculation or rule.
Explain unusual entries
- Group related exceptions and summarize their source records for the reviewer.
- Distinguish known seasonality and approved adjustments from unexplained changes.
Checkpoint: Narratives cite entry IDs and state uncertainty.
Assign the investigation queue
- Map each exception to its accounting owner and required evidence.
- Carry forward unresolved items while retaining stable keys for duplicate prevention.
Checkpoint: The close team has a prioritized, nonduplicated queue.
Capture resolutions
- Record the owner's explanation, evidence and any authorized correction reference.
- Re-run affected checks after corrections and leave unverified items open.
Checkpoint: Every closed exception has a supported resolution.
Decisions along the way
Export totals do not reconcile
Stop account-level conclusions and repair the extract first.
Unusual entry is an approved adjustment
Record the approval reference and close only the matching exception.
Materiality rule is absent
Report values without assigning a materiality severity.
What you will produce
close-exceptions.csv
CSVA working close queue preserving investigation evidence.
Include: exception_id · account · entry_ids · rule · amount · owner · status · resolution
close-coverage.md
MarkdownThe controller can distinguish clean results from untested areas.
Include: required_sources · received_sources · cutoffs · unassessed_checks
Check the result
- Reconcile each source's row count and totals before comparing accounts.
- Verify deterministic checks reproduce the listed exceptions.
- Sample unflagged entries to detect missed cases.
- Confirm resolved items retain evidence and any journal reference.
When something goes wrong
Late postings change the period
Create a new snapshot and compare changes; do not overwrite the prior close record.
Owner mapping is missing
Place the item in an unassigned queue for the controller.
AI explanation contradicts an entry
Discard the explanation, keep the raw exception and request reviewer interpretation.
A worked example
SYNTHETIC EXAMPLE · NOT A REPORTED BUSINESS RESULT
Request
Synthetic: prepare the April close queue.
Inputs
Ledger entry J7 has two identical 2,400 USD debits with the same supplier reference. Account A9 reconciliation status is blank. An unusual 8,000 USD adjustment J8 has a valid controller approval.
Expected result
Synthetic result: flag J7 as potential duplicate, A9 as reconciliation unknown, and retain J8's approval evidence. Do not propose deleting J7 or treating A9 as reconciled.
Run a small pilot
Start small
Run beside one completed close using known exceptions as a comparison set.
Record the baseline
Record analyst review time, duplicate investigations and missed material items.
Agree success criteria
Agree coverage and false-negative criteria with the controller before operational reliance.
Stop or revise
Stop automated closure if source totals or correction references cannot be reconciled.
Keep the scope clear
- Exception preparation does not authorize journal posting or close sign-off.
- An anomaly is a review prompt, not a finding of misconduct or incorrect accounting.
Check before using
The controller verifies exceptions and authorizes corrections before posting.
Business activities
- Manage finances and accounting
FROM A RECIPE TO YOUR WORKFLOW
Bring the company context together.
This workflow needs reliable inputs before it needs a better prompt. Moat8 Vault brings working information into a readable archive for the AI tools you use.
For this recipe: Period accounting records · Close checklist and exception rules
Start with one recurring task, agree which sources to use, and review the first result with your team.
Discuss this workflow