Score every order in milliseconds
Machine-learning risk scores with plain-language reasons, and thresholds you control, so good orders flow and bad ones stop.
Why it matters
Fast
Scores return before the shopper notices.
Explainable
Every score lists the signals that moved it.
Your thresholds
Decide where approve, review and decline begin.
What's inside
Scoring
- Device, behaviour and network signals
- Market-aware models
- Reasons attached to every score
42 msmedian fraud-scoring time1
How it works
- 1
Collect signals
Checkout components gather device and behaviour signals.
- 2
Score
The model scores the order and explains why.
- 3
Decide
Your thresholds approve, send to 3DS, review or decline.
- 4
Learn
Chargebacks and review outcomes feed back into the model.
Try it
Set your own thresholds
Move the order's risk score and the two thresholds to see what happens to it.
Challenge with 3DS
Medium risk. The shopper confirms in their banking app; liability shifts to the issuer if they pass.
Signals behind this score
- Device seen on 6 earlier orders−12
- Billing and shipping countries match−8
- Order value close to this shopper's usual range−5
- Email domain created recently+6
Interactive demo with illustrative data.
curl https://api.sandbox.international-orders.com/v1/orders/ord_8KQ2/risk \
-H "Authorization: Bearer $IO_SECRET_KEY"Customer story
Lumen Games
11
new markets in one quarter · Illustrative
Local wallets were the unlock in Korea and Japan. Fraud Shield kept chargebacks flat while we grew.
Aiko Tanaka, Payments Lead Read the story: Lumen Games
Questions about Fraud Shield
Put Fraud Shield to work
Test it in the sandbox today, or talk to us about your markets.