Fraud & Risk

Fraud & Risk surfaces things worth a second look: signs that a statement may not be internally consistent, and per-transaction risk signals you can pull up on demand.

Signals, not verdicts

Nothing on this page determines that fraud has occurred. A score, a level or a finding is a prompt to review — evidence to check, not a conclusion. See Notes and limits.

What it shows

Panel	What it contains
Integrity score	A score out of 100 for the statement data in view, with an accompanying risk level of LOW, MEDIUM or HIGH
Statement integrity	A short written summary, plus a list of findings. Each finding has a title, an explanation, how many times it occurred, its own severity, and example values
Benford's-Law digit distribution	For each leading digit 1–9, the share of amounts that start with it, shown against the naturally expected share, with a note on whether the distribution looks natural or deviates enough to be worth verifying
Per-transaction risk	A table of your transactions — date, description or merchant, and amount — with a Check action per row

When no integrity issues are found, the panel says so explicitly rather than showing an empty list.

Features

Feature	What it does
Statement-integrity checks	Looks across the statement data for internal inconsistencies — including balance continuity, duplicates, round-number patterns and digit distribution — and reports what it finds
Findings with examples	Every finding carries a count and sample values so you can go and look at the actual rows
Benford comparison	Shows the observed leading-digit distribution against the expected one, so an unusual shape is visible at a glance
On-demand transaction check	Choose Check on any row to retrieve that transaction's risk level, its score and the reasons behind it
Reasons	Each checked transaction lists the specific reasons that contributed, shown as labels next to the row

How to use it

1. Open Fraud & Risk.
2. Read the integrity score and risk level for an overall sense of the data in view.
3. Work through the findings. For each one, use the count and the examples to judge whether it looks like a genuine problem or an artefact of how the account is used.
4. Glance at the Benford panel. A distribution flagged as deviating is a reason to verify, not a conclusion.
5. In the transaction table, choose Check on any row you want assessed. The level, score and reasons appear on that row.
6. Follow up on anything that looks wrong using your own records — the original statement, the merchant, or your bank.

Notes and limits

• These are risk signals, not determinations of fraud. A HIGH level or a low integrity score means "review this", never "this is fraudulent". Do not treat a score as proof, and do not act against a person or a supplier on the strength of one.
• Equally, a LOW level and a clean findings list are not a guarantee that nothing is wrong. Absence of a signal is not evidence of absence.
• Findings describe patterns in the data you uploaded. Legitimate activity can produce them — regular round-number transfers, repeated identical charges from one merchant, or a genuinely small sample of transactions can all move the picture.
• The Benford comparison is only meaningful on a reasonable volume of amounts. Read it with the sample size in mind.
• Per-transaction risk is retrieved when you choose Check on a row. Rows you have not checked simply have no assessment yet.
• The transaction table shows a recent page of transactions rather than your entire history. Use Transactions to search and filter the full set.
• Integrity analysis covers the statement data available for the period in view. Uploading more statements — or correcting categorisation — can change the result.