Free online tool
Screen a ledger for Benford and anomaly flags
Paste ledger amounts or drop a CSV and this free tool screens them in your browser: a Benford first-digit test with chi-square and MAD, duplicate amounts, round-number clustering, sequence gaps, and the five largest entries, rolled into a Low, Review, or High risk score. Nothing is uploaded and nothing is stored. A flag means “look here” — it is a screening aid, not proof of fraud, and every flagged pattern needs a human to check the underlying entries.
Quick answer
Quick answer
Paste ledger amounts or drop a CSV and this free tool screens them in your browser: a Benford first-digit test with chi-square and MAD, duplicate amounts, round-number clustering, sequence gaps, and the five largest entries, rolled into a Low, Review, or High risk score. Nothing is uploaded — and a flag means review, not fraud.
Open the interactive Audit Ledger Anomaly Screener
The interactive tool loads in the page. Enable JavaScript to enter data, process content, and download results.
Worked example — the built-in sample
The built-in sample ledger: 60 entries, entry numbers 1–61 with 37 missing.
| Entries screened | 60 — total $215,638.96 |
|---|---|
| Benford MAD | 0.008 — Acceptable conformity |
| Chi-square | 0.41 — below the 15.51 critical value |
| Duplicates | 3 amounts across 7 entries — $1,250.00 × 3, $2,400.00 × 2, $499.99 × 2 |
| Round numbers | 10 of 60 (16.7%), of which 4 are round thousands |
| Sequence gap | Entry 37 missing |
| Overall risk | Review — 2 points |
Benford first digits — the sample ledger
| First digit | Observed | Expected (Benford) | Difference |
|---|---|---|---|
| 1 | 16 (26.7%) | 30.1% | -3.4pp |
| 2 | 11 (18.3%) | 17.6% | +0.7pp |
| 3 | 8 (13.3%) | 12.5% | +0.8pp |
| 4 | 6 (10.0%) | 9.7% | +0.3pp |
| 5 | 5 (8.3%) | 7.9% | +0.4pp |
| 6 | 4 (6.7%) | 6.7% | 0.0pp |
| 7 | 4 (6.7%) | 5.8% | +0.9pp |
| 8 | 3 (5.0%) | 5.1% | -0.1pp |
| 9 | 3 (5.0%) | 4.6% | +0.4pp |
MAD conformity ranges (Nigrini)
| MAD range | Conformity | What it means |
|---|---|---|
| 0.000–0.006 | Close conformity | First digits track Benford's law closely |
| 0.006–0.012 | Acceptable conformity | Normal variation for a real ledger |
| 0.012–0.015 | Marginally acceptable conformity | Worth a closer look at the deviating digits |
| Above 0.015 | Nonconformity | First digits diverge — review the entries behind the gap |
About this tool
About the Audit Ledger Anomaly Screener
Paste ledger amounts or drop a CSV and this free tool screens them in your browser: a Benford first-digit test with chi-square and MAD, duplicate amounts, round-number clustering, sequence gaps, and the five largest entries, rolled into a Low, Review, or High risk score. Nothing is uploaded and nothing is stored. A flag means “look here” — it is a screening aid, not proof of fraud, and every flagged pattern needs a human to check the underlying entries.
Detailed guide
How to get useful results from Audit Ledger Anomaly Screener
What Benford's law says about ledgers
Benford's law describes the first digits of numbers that span several orders of magnitude. In such data, 1 is the most common first digit at about 30.1%, 2 follows at 17.6%, and the shares fall steadily to 4.6% for 9. Real ledgers — sales, expenses, invoices, payments — usually follow this pattern, because the amounts are the product of many unrelated prices and quantities.
The pattern holds only when the numbers are free to range widely. It breaks down for assigned numbers (invoice numbers, cheque numbers), for prices fixed by a list, and for amounts constrained by a minimum or maximum. That is why the tool shows the full per-digit table instead of a single pass/fail: you can see which digits drift and judge whether the ledger's own rules explain it.
Why invented numbers fail the first-digit test
People inventing numbers tend to spread first digits too evenly. Asked to make up figures, most of us reach for 5s, 6s, and 7s and underuse 1s — the opposite of Benford's distribution. Auditors have used this as a first pass for decades: a ledger whose first digits sit far from Benford's law is not proven false, but it is worth opening.
The tool quantifies the drift two ways. The chi-square statistic compares observed and expected counts across all nine digits; above the critical value of 15.51 (8 degrees of freedom, 5% level), the gap is unlikely to be chance. The mean absolute deviation (MAD) averages the per-digit gaps, and Nigrini's published ranges label it: up to 0.006 is close conformity, 0.006–0.012 acceptable, 0.012–0.015 marginally acceptable, above 0.015 nonconformity.
Duplicates and round numbers
Benford's test cannot see repeats, so the screen checks them separately. Every amount that appears more than once is listed with its count, and the share of entries involved is shown. Repeats are normal for fixed prices, subscriptions, and retainers — they matter most when the same round amount repeats with no reason, or when one value appears far more often than the business would produce.
Round-number clustering counts the amounts that are whole multiples of $100, and how many of those are round thousands. Estimates, deposits, and fees are usually round, so some round numbers are expected. A ledger where a fifth or more of the amounts end in 00 is unusual enough to ask about: real spending is rarely that tidy. The built-in sample has 10 round numbers out of 60 (16.7%), which is inside the normal range.
Sequence gaps and missing entries
When the ledger carries entry numbers, a gap is a question in itself. The tool lists every missing number between the lowest and highest entry, so a single deleted or voided entry shows up as plainly as a run of missing ones. A gap can be innocent — a voided invoice, a numbering mistake — but it should have an explanation attached to it.
To check gaps, upload or paste a two-column CSV with an entry,amount header, or tick the “First column is an entry / sequence number” box for a headerless paste. A single column of amounts has no numbers to check, so the tool says so instead of guessing. The built-in sample has one gap: entry 37 is missing from numbers 1–61.
How the risk score works
The risk score is a triage aid, built from five checks. A MAD above 0.015 scores 2 points, 0.012–0.015 scores 1; a significant chi-square scores 1; duplicate rows above 5% of entries score 1; round numbers above 20% score 1; any sequence gap scores 1. Totals of 0–1 are Low, 2–3 are Review, and 4 or more are High.
The score deliberately caps each check, so no single pattern can push a ledger to High on its own. Low means nothing stood out — keep the usual checks. Review means a human should look at the flagged items, starting with the largest ones. High means look first. None of the three is a finding: the score ranks ledgers for review, it does not judge them.
Reading the sample ledger
The built-in sample is a 60-entry ledger with entry numbers 1–61 and entry 37 missing. Its first digits run 16, 11, 8, 6, 5, 4, 4, 3, 3 — close to Benford's expected 18.1, 10.6, 7.5, 5.8, 4.8, 4.0, 3.5, 3.1, 2.7 — giving a MAD of 0.008 (acceptable conformity) and a chi-square of 0.41, far below the 15.51 critical value.
The sample still earns a Review: $1,250.00 appears 3 times, $2,400.00 and $499.99 twice each — 3 duplicate amounts across 7 entries, 11.7% of the ledger — and entry 37 is missing. Those two points are exactly the kind of small, checkable findings a first-pass screen is for: each one takes a minute to verify against the books.
What this screen cannot prove
This screen reads amounts only. It cannot see dates, vendors, or descriptions, so it cannot tell a genuine repeat from a copied one, and it cannot tell whether a round number is an estimate or a fabrication. Small ledgers — under about 100 entries — make the Benford test noisy, so a small sample can flag by chance alone.
Enterprise audit-analytics platforms bundle this kind of screening into contracts that cost thousands of dollars a year, with case management and data connectors on top. This tool gives you the core first-pass checks — Benford, duplicates, round numbers, gaps — free in your browser, without the rest of the platform. What it cannot give you is a verdict: only a person who knows the business can say whether a flag is fraud, a mistake, or nothing at all.
Quick guide
How to use this tool
- 01Paste ledger amounts one per line, or drop a .csv/.txt file — with an entry,amount header for sequence checks.
- 02Tick “First column is an entry / sequence number” if your paste has two columns and no header.
- 03Press Screen ledger and read the risk score first — Low, Review, or High.
- 04Open the Benford table to see which first digits drift, then check duplicates, round numbers, and gaps.
- 05Download the results CSV to keep the numbers, and review the flagged entries in your own books.
Capabilities
Features and supported formats
- Benford first-digit test with observed vs expected shares per digit, chi-square statistic, and MAD with Nigrini conformity ranges.
- Duplicate-amount finder listing every repeated value with its count and the share of entries involved.
- Round-number clustering: the share of amounts that are whole multiples of $100, and how many are round thousands.
- Sequence-gap check for two-column entry,amount CSVs, listing every missing entry number.
- Largest-five table plus entry count, total, and mean — so the biggest values get eyes on them first.
- One-click sample ledger, clear, and CSV export of the full results — all in the browser, nothing uploaded.
Supported formats
Input: ledger amounts pasted one per line, or a .csv/.txt file — one amount per line, or a two-column entry,amount CSV — up to 20,000 entries. Output: Benford table with chi-square and MAD, duplicates, round-number share, sequence gaps, largest entries, and a Low / Review / High risk score, with CSV export of the results.
Answers
Frequently asked questions
Does the Audit Ledger Anomaly Screener upload my ledger?
No. Amounts are pasted or read with the File API and screened only in your browser tab. Nothing is sent to a server, nothing is stored, and no amount ever leaves your device.
What is Benford's law?
Benford's law says that in numbers spanning several orders of magnitude, 1 is the most common first digit (about 30.1%) and 9 the least (about 4.6%). Real ledgers usually follow it; invented numbers tend to spread first digits too evenly.
What does the MAD score mean?
MAD is the mean absolute deviation — the average gap between your observed first-digit shares and Benford's expected shares. Nigrini's ranges: up to 0.006 is close conformity, 0.006–0.012 acceptable, 0.012–0.015 marginally acceptable, above 0.015 nonconformity.
What is the chi-square value?
Chi-square compares your observed first-digit counts against Benford's expected counts across all nine digits. Above the critical value of 15.51 (8 degrees of freedom, 5% level), the gap is unlikely to be chance — but it is still only a flag, not a finding.
How many entries do I need for the Benford test?
More is better. Under about 100 entries the test is noisy and can flag by chance. The tool still runs on small ledgers — it shows the per-digit table so you can see whether the drift is one digit or all of them — but treat small-sample flags lightly.
What counts as a duplicate amount?
Any exact value that appears more than once, to two decimal places. The tool lists each repeated amount with its count. Repeats are normal for fixed prices and subscriptions; they matter most when a round amount repeats with no business reason.
Why do round numbers matter?
Real spending is rarely tidy. Estimates, deposits, and fees are usually round, so some round numbers are expected — but when a fifth or more of a ledger ends in 00, it is worth asking why. The tool counts whole multiples of $100 and round thousands separately.
How do sequence gaps work?
If your data has an entry-number column, the tool lists every missing number between the lowest and highest entry. Upload a two-column entry,amount CSV, or tick the sequence-number box for a headerless paste. A gap can be a voided entry or a numbering mistake — check what that number was.
What does the risk score mean?
Low means nothing stood out, Review means a human should look at the flagged items, High means look first. The score is built from five checks (MAD, chi-square, duplicates, round numbers, gaps) and ranks ledgers for review — it is not a verdict.
Can this screen prove fraud?
No, and no screening tool can. A flag means “look here”, not “this is fraud”. Real ledgers break Benford's law for honest reasons — price lists, minimum charges, and assigned numbers all skew first digits. Every flag needs a human to check the underlying entries.
What file formats work?
Paste amounts one per line (with or without $ and thousand separators), or drop a .csv or .txt file — one amount per line, or a two-column entry,amount CSV. Files up to 5 MB, pastes up to 20,000 entries.
Can I use it on my phone?
Yes. The paste box and file picker work in mobile browsers, and the whole screen runs in the tab. Paste a column of amounts from your ledger app and the results appear in seconds.