Establish the baselines
Disbursement, card and payroll history is loaded to establish the entity's own patterns. Thresholds are drawn from the entity's policies.
Fraud Detection
Detection ranks. Rules articulate. People decide. Systems that skip the middle step produce findings nobody can defend.
Anomaly detection
What it does
Fraud Detection is a triage tool for post-payment review. It ranks transactions by how unusual they are relative to the entity's own history and its peer groups, and pairs every ranking with the policy provision the item potentially implicates.
It does not block payments. Pre-payment intervention is a materially different risk posture — it interacts with prompt-payment obligations and requires an override procedure that is auditable and fast — and entities should establish review discipline before considering it.
Interface mockup. Values shown are illustrative and do not represent any entity.
Capabilities
What is included
- Split-transaction detection. Purchases grouped by cardholder, merchant and rolling window, flagged where the group exceeds a single-transaction limit each component falls below.
- Vendor master monitoring. Duplicate detection on name, tax identifier, bank details and address; employee–vendor overlap; dormancy; and bank detail change alerts.
- Peer-group anomaly scoring. Cardholders and cost centres compared against role- and department-based peers, or against their own history where no peer group exists.
- Payroll exception review. Ghost employee indicators, unusual rate changes, overtime concentration, and payments after separation date.
- Digit and duplicate analytics. Digit distribution and repeated-value testing used to select samples, with the documented exclusions the tests require.
- Override logging. Every disposition and reversal recorded with a reason code, producing the override-rate data that shows whether the tool is working.
Implementation
How a deployment runs
Typical first deployment reaches steady state within one to two quarters, depending on the state of the source data.
Generate the ranked worklist
Each cycle produces an ordered queue with the flagged values and the provision each item potentially implicates.
Review and dispose
Reviewers record a disposition with a reason. Escalation paths route around implicated supervisors.
Recalibrate
Rules generating alerts but no confirmed exceptions are revised or retired, so reviewer attention is not eroded by noise.
Limits
What Fraud Detection does not do
We publish this section for every tool. It is the section we would want to read first.
- Anomaly detection identifies deviation from history, not deviation from policy. Where an improper practice has been routine, it becomes the baseline.
- Output is a review queue, not a finding. Nothing produced by the system constitutes evidence of impropriety.
- RIGFOA does not conduct investigations, does not contact subjects of review, and does not make referrals.
Talk to us about your oversight programme
Walk through the platform with your own chart of accounts, or start with the research library. Both routes are free to begin.