Fraud Scoring
Score a transaction's risk in under 50 milliseconds.
About this API
Combines device fingerprinting, IP reputation, email and phone age, velocity rules, and a gradient-boosted risk model into a single 0-100 score with human-readable reason codes you can show a review team.
Common use cases
Quickstart
Every Zephiel API uses the same base URL, the same key header, and the same error envelope.
curl -X GET "https://zephiel.com/api/v1/fraud-scoring/score" \
-H "X-Zephiel-Key: $ZEPHIEL_API_KEY" \
-H "Accept: application/json"Example response
{
"score": 78,
"decision": "review",
"reasons": [
"email_domain_age_under_30_days",
"ip_country_mismatch_billing"
],
"latency_ms": 41
}/v1/fraud-scoring/scoreScore a transaction or signup event/v1/fraud-scoring/feedbackReport the true outcome to retrain/v1/fraud-scoring/rulesList active velocity rulesAll endpoints return JSON and accept the X-Zephiel-Key header. Rate limits are enforced per plan and reported in X-RateLimit-Remaining.
Plans can be changed or cancelled at any time. Overage is charged per 1,000 requests rather than hard-blocking your traffic.
Free
100 requests/mo
5 req/min
Free
- JSON output format — included
- HTTPS encryption — included
- API key authentication — included
- Standard endpoints — included
- Community support — included
- CSV output format — not included
- Email support — not included
- Usage analytics and alerts — not included
- 99.9% uptime SLA — not included
- Webhooks — not included
- Priority support — not included
- Custom rate limits — not included
- 99.99% uptime SLA — not included
- Dedicated success manager — not included
- SSO / SAML — not included
- Custom SLA, DPA, and BAA — not included
- Private deployment option — not included
- Volume pricing — not included
Starter
10,000 requests/mo
60 req/min
$45/month
- JSON output format — included
- HTTPS encryption — included
- API key authentication — included
- Standard endpoints — included
- Community support — included
- CSV output format — included
- Email support — included
- Usage analytics and alerts — included
- 99.9% uptime SLA — included
- Webhooks — not included
- Priority support — not included
- Custom rate limits — not included
- 99.99% uptime SLA — not included
- Dedicated success manager — not included
- SSO / SAML — not included
- Custom SLA, DPA, and BAA — not included
- Private deployment option — not included
- Volume pricing — not included
Most popular
Pro
250,000 requests/mo
600 req/min
$180/month
- JSON output format — included
- HTTPS encryption — included
- API key authentication — included
- Standard endpoints — included
- Community support — included
- CSV output format — included
- Email support — included
- Usage analytics and alerts — included
- 99.9% uptime SLA — included
- Webhooks — included
- Priority support — included
- Custom rate limits — included
- 99.99% uptime SLA — included
- Dedicated success manager — not included
- SSO / SAML — not included
- Custom SLA, DPA, and BAA — not included
- Private deployment option — not included
- Volume pricing — not included
Enterprise
Unlimited requests
Custom
$630/month
- JSON output format — included
- HTTPS encryption — included
- API key authentication — included
- Standard endpoints — included
- Community support — included
- CSV output format — included
- Email support — included
- Usage analytics and alerts — included
- 99.9% uptime SLA — included
- Webhooks — included
- Priority support — included
- Custom rate limits — included
- 99.99% uptime SLA — included
- Dedicated success manager — included
- SSO / SAML — included
- Custom SLA, DPA, and BAA — included
- Private deployment option — included
- Volume pricing — included
4.8
12 reviews
Grace Mwangi
Head of Product
Held up under real load
We push Fraud Scoring hard during peak and latency has stayed flat. No throttling surprises, no silent degradation — the numbers on the listing matched what we measured.
Mei Chen
Integration Engineer
Good coverage for fraud
We checked Fraud Scoring against our own fraud reference data before committing. Coverage was better than the incumbent we were paying four times as much for.
Peter Osei
DevOps Lead
Documentation matched reality
Every sample in the Fraud Scoring docs ran unmodified. That sounds like a low bar until you have integrated something where it is not.
Rachel Kimani
Head of Integrations
Now handles account opening too
We adopted Fraud Scoring for checkout risk review and ended up using it for account opening as well. Same key, no new contract, no second integration.
Chidi Nwosu
Head of Engineering
Support answered properly
Asked Meridian AI about an edge case in Fraud Scoring and got an explanation from someone who had clearly read the code, not a link to the FAQ.
James O'Connor
Technical Director
Quietly reliable
Fraud Scoring has not surprised us once in eight months of production traffic. That is the highest compliment I can pay a dependency.
Aisha Suleiman
Digital Lead
One key made the difference
Adding Fraud Scoring took ten minutes because the credential already existed. The vendor onboarding we skipped would have taken longer than the integration.
Anna Kowalski
Senior Developer
Solved Checkout risk review for us
We brought Fraud Scoring in for Checkout risk review and it has not needed revisiting since. The response shape matched the documentation on the first call, which is rarer than it should be.