Full reasoning chain
Every conclusion carries its reasoning and a citation to the source document, with agent actions and human overrides recorded in sequence.
Solution · Merchant Fraud
Transaction laundering, bust-out merchants, collusive refunds and chargeback abuse. The signal is almost never in one transaction. It is in the portfolio, over weeks, across merchants that are not supposed to be related.
The work
A merchant case is a worked example of the problem: every individual transaction is defensible, the population is enormous, and the finding has to hold up against a merchant who will dispute it and a scheme that will move to the next acquirer.
A monitoring rule, a chargeback threshold, an acquirer enquiry or a referral opens the file — and the scope is set.
The merchant's own history and its peer set are established, so the deviation is measured against something rather than a fixed threshold.
Refund and chargeback patterns, volume shifts, ticket-size drift, and processing that does not match the stated business are tested.
Ownership, settlement accounts, bank details and addresses are mapped across the portfolio to surface merchants that are not supposed to be linked.
The determination, the evidence behind each finding, and the actions your policy mandates — hold, terminate, refer or report.
Where the agents run
The same three agents run this case type as run every other. What differs is the policy behind them — the plan template, the dimensions scored, the tools available, and the report that comes out.
One merchant rarely looks wrong enough to act on. Five merchants sharing a settlement account, a director and the same chargeback curve do. RiskPulse tests each merchant against its own history and its peers, then maps the relationships across the book — so what reaches an investigator is the cluster and the evidence for it, not the single merchant that happened to trip a threshold.
Policy for this case type
Nothing on this page is fixed in the product. Every element below is a policy setting you define, which is why the same platform runs this case type and the others.
Controls
The same governance and quality controls apply to this case type as to every other — because an examiner does not lower the standard for the harder files.
Every conclusion carries its reasoning and a citation to the source document, with agent actions and human overrides recorded in sequence.
A sampling methodology sits behind this case type, so your quality team can assess a representative sample rather than reviewing at random.
A separate set of agents assesses case quality end to end against your methodology — a second opinion on the work.
See it on your own file
Pick a real file or a closed case and see how RiskPulse plans it, evidences it, and writes it up under your policy.
Request Demo