Ledger files as they come
Import a FEC, tab or pipe separated, or a ledger export in CSV, TSV or text, up to 2 GB. The general ledger checks need a FEC with its mandatory columns. You confirm the encoding before the import.
Structure findings on import
The import checks the FEC structure: column order, gaps and duplicates in entry numbers, unbalanced entries, missing amounts and missing validation dates.
Journal entry tests
Unbalanced entries, duplicates, weekend and French public holiday postings from 2015 to 2030, round amounts, amounts at or above your threshold, rare account pairs and Benford’s law.
A journal of every step
Each step records its settings, the engine version and fingerprints of its inputs and output. Each entry is chained to the one before it.
Replay up to any step
Replay runs the journal again from the first entry and gives a verdict per step: identical or different. A step whose operation version was retired, and the steps that use its result, are reported as retired and not replayed. The replay is added to the journal and overwrites nothing.
Results in Excel
In the Excel add-in, bring a result into a sheet, up to 10,000 rows. You can also run the checks on a sheet, as a local run kept out of the journal.
What the analytics cover
General ledger checks
One run covers the whole ledger. You set the closing date, a large amount threshold and a term to search for in entry labels.
- Unbalanced entries, duplicate lines and gaps in entry numbers.
- Entries dated or validated on a weekend or a French public holiday. The holiday table covers 2015 to 2030, and a date outside those years stops the run.
- Round amounts, amounts ending in 999 and amounts at or above your threshold.
- Rare debit and credit account pairs, and accounts used on 3 lines or fewer.
- Entries around the closing date reversed after it, and validations after closing.
- Benford’s law on the first digit, the first two digits, the second digit and the last two digits.
- Balances per account, per journal and per month.
The engine computes, the assistant prepares
Every result comes from the analytics engine, with exact decimals and a stable row order. The assistant can propose a run, and nothing starts until a member with full access approves it.
Verify the chain
The check finds a journal entry that was changed, removed, inserted or moved, a broken link between entries, or a journal cut short.
Analysis projects
Each analysis lives in a project with its own members, with full access or read only. Field statistics give each column its counts, missing and distinct values, minimum and maximum. Number columns add the sum and mean, text columns their empty texts, and date columns counts by weekday and month.
Export and retention
Export a closed project as an archive: its sources, the journal, the results as CSV and a manifest. Ask for a new archive at least 24 hours before deletion. A project’s sources, journal and results are deleted at the latest 30 days after it is closed.
- A project with no activity for 12 months gets a warning. It closes only if it stays inactive for the next 30 days, and any activity clears the warning.
From the FEC to tested results
Import the ledger
Create a project and upload the FEC. Confirm the encoding and read the import findings.
Run the checks
Set the closing date, the threshold and the search term, then launch the general ledger checks.
Review and replay
Open each result, verify the journal chain, replay it and bring the results you keep into Excel.
Questions about data analytics
- Which files can I import?
- CSV, TSV or text exports up to 2 GB, FEC files included. XML FEC files and Excel workbooks are not read as sources. From Excel, send a range through the add-in instead.
- Does the AI compute the results?
- No. The analytics engine computes every result. The assistant can propose a run and read a summary of at most 20 rows, and a member with full access approves each run.