Fraud caught, refunds handled, reputation defended — on autopilot.
Melissa scores every order against 35+ signals, issues refunds through a single guard-railed gate, and watches your reviews around the clock. The expensive mistakes get caught before they cost you.
Fraud scoring
35+ signals, three clear decisions
Every order gets a 0–100 risk score across payment, identity, address, and behavioral signals — mapped to LOW (auto-approve), MEDIUM (review), and HIGH (reject). Every score shows the exact rules that fired, with weights.
- Explainable rules: CVV/AVS, velocity, reshipper, disposable email
- Three decision bands with seasonal relaxation
- Velocity detection: card/email cycling, drop addresses
- Dollar exposure by risk tier
Fraud scoring · last 24h
0–29
Auto-approve + capture
30–69
Manual review — no capture
70–100
Auto-reject (surfaced only)
| Score | Customer | Order | Matched rules |
|---|---|---|---|
78HIGH | MC Marcus Chen m.chen@tempmail.io | #PO-2847$145.99 | Reshipper address+40Velocity email/cards+35CVV mismatch+25 |
85HIGH | U Unknown disposable@mailinator.com | #PO-2841$175.00 | Disposable email+25First + rush + cross-state+25PO box / CMRA+20 |
52MEDIUM | JW Jennifer Wong jwong@gmail.com | #PO-2843$89.50 | Card multi-city 24h+35AVS zip mismatch+25Name mismatch+10 |
22LOW | RD Robert Diaz rdiaz92@gmail.com | #PO-2844$62.75 | Classifier signal+7Rush first order+5Generic gift message+5 |
Scores are model-calibrated across 35+ signals — not a raw sum of rule weights.
Automatic refunds
Refunds with guardrails, not blank checks
Every refund — from AI replies, the bank queue, or one-click operator actions — flows through one serializable gate with daily-$, velocity, monthly-ratio, and per-refund caps. Decisions key off live gateway truth: void before settlement, refund after.
- Single gate for every refund channel (AI, email, Slack, operator)
- Daily, hourly, monthly & per-refund caps
- AI pacing plan — a daily refund strategy that stays within your caps
- Void vs. refund based on live Authorize.Net state
- Full audit trail + CSV export
Budget guardrails
52% of daily cap used
Monthly ratio
2.8%
Count
18 / 50
Per-refund max
$300
Nearest cap
daily_amount
Refunds & bank queue
| Order | Customer | Amount | Auth.net | Status |
|---|---|---|---|---|
| #PO-4427 | Sarah Chen | $89.50 | auto | sent |
| #PO-4421 | Marcus Johnson | $62.00 | bank | sent |
| #PO-4419 | Lisa Rodriguez | $35.50 | auto | queued |
| #PO-4410 | David Kim | $124.00 | operator | sent |
| #PO-4432 | Patricia Nelson | $152.00 | Settled | |
| #PO-4428 | Robert Chang | $96.50 | Captured |
Reputation watch
Reviews triaged before they snowball
Melissa ingests Trustpilot and BBB reviews, classifies sentiment and risk, matches them to orders, and tracks SLA deadlines — auto-escalating anything left unanswered. A 1-star scam claim hits your desk in minutes, not days.
- Auto-ingest Trustpilot & BBB
- AI sentiment, category & risk scoring
- Order matching + SLA escalation
- Suggested replies, ready to publish
Reputation center
| Reviewer | Rating | Risk | Status | SLA |
|---|---|---|---|---|
TB Thomas Bradshaw BBB · bad | 95 | escalated | breached | |
MJ Marcus Johnson BBB · bad | 85 | escalated | 8h left | |
JW Jennifer Wu BBB · bad | 70 | investigating | today | |
SM Sarah Mitchell Trustpilot · good | 0 | testimonial | — | |
DC David Chen Trustpilot · good | 0 | no reply needed | — |
Catch the costly mistakes before they cost you.
Book a demo and see how Melissa scores fraud, guards every refund, and defends your reputation — automatically.