L
LoyaltyDocs

Approvals & Fraud Review

The review queue — conflicts, fraud signals, attribution and milestone approvals — the decisions a reviewer can make, and the fraud rules behind it

Whenever the platform is not sure a referral should pay, it asks a person. Every such question is a review in one queue — Referrals → Review — with one of two answers: approve or reject.

What Lands in the Queue

Review

Raised when

Queue view

Two-referrer conflict

Two referrers claim the same referee — the second capture opens the review and the claims wait

Conflicts

Self referral, Family match, Shared contact, Capture velocity, Blacklisted referrer, Fraud signal

A fraud rule fired on capture

Fraud

Attribution approval

The capture came through a channel set to Needs approval, or the program holds captures until verified

Approvals

Milestone approval

A program set (over the API) to approve each milestone had a milestone fire

Approvals

Each card shows the referee, the referrer(s) involved, the channel, the reason and its severity, and — for a velocity rule — the evidence: how many captures in the window against the limit.

The Two Decisions

Decision

Conflict

Fraud or attribution

Milestone

Approve — Pick as winner on a conflict

Accepts the referral the review is for; every other competing claim for that referee is cleared

Accepts the referral; its first stage fires

Credits the held rewards

Reject (reason required)

Rejects every competing claim

Rejects the referral and voids anything held

Forfeits the held rewards

Only the referrer side is affected by a decision — the referee keeps rewards they earned through their own events. A review is decided once; a second decision on the same review is refused. Every decision is audited with its note, and resolved items move to Recently resolved with a Matured or Reversed chip. Administrators can decide any review; other agents can only when the program's Agents can release held rewards setting allows it.

Approve the right one in a conflict

A conflict card lists every claiming referrer. Pick the one that should win — the platform clears the rest for you.

Fraud Rules

Fraud checks run at capture, before any acceptance. They are tenant-wide, under Settings → Fraud rules, and off until you switch them on:

Setting

What you choose

Default

Enable fraud detection

On / off

Off

Referral captures / referrer and Capture window (days)

The most referees one referrer may capture within the window; blank for no limit

No limit

Block self-referrals

Refuse a referee whose phone belongs to the referrer

Off

Action

Flag — the referral is held for review / Freeze member until reviewed — the member's rewards are frozen as well

Flag

Severity

Low / Medium / High — recorded on every flag for triage

Medium

Whitelist

Phones that fraud rules never flag

—

Blacklist

Phones whose captures are always sent to review

—

The same page lists the open fraud holds with Approve and Reject, so fraud can be worked without leaving settings. The two-referrer check runs on every capture whether or not fraud detection is on.

Recorded, not yet enforced

The fraud page also offers *OTP failures before lockout*, *Flag shared device* and *Flag round-number bills over a threshold*. These are not read by the engine today; the rules above are.

For developers

Decide a review with `POST /api/v1/loyalty/referral/reviews/{review_uuid}/resolve` and a body of `{ "decision": "APPROVE" | "REJECT", "resolution_note": "…" }`. Fraud rules are written with `PUT /api/v1/loyalty/fraud/config`. The queue itself, with its type and status filters, streams to the app over its live connection and refreshes as decisions are made.

Was this page helpful?

Last updated on

On this page