/01
Fraud score
Every triggered rule, summed into one number — the primary signal for approve, review or decline.
Atna Scoring Engine
Decisioning intelligence for every transaction.
Atna's Scoring Engine evaluates every signal in a transaction, turning your rules and machine-learning models into one verdict — approve, review, or decline — with absolute, auditable precision.
scoring_engine · live decisioninglive
0fraud score
verdict
REVIEW
4 rules triggered · manual checkApprove
8
whitelist
Review
64
3 signals
Decline
100
blacklist
applied rules · txn 0xA7F2-4471
Risky IP — datacenter / proxyR-204+18
Disposable email domainR-118+12
No online profiles foundR-061+7
Email is not deliverableR-044+4
Scoring Intelligence
Navigating the risk landscape.
Sophisticated fraud requires layered precision. The engine reads every vector a single check would miss, then weighs each one into the decision.
/02
AI Insights score
An independent 0–100 probability from the model, catching complex patterns rules alone miss.
/03
Email & phone network
Risk from signals tied to the email or phone, enriched with global consortium fraud history.
/04
IP intelligence
VPN, TOR, datacenter and proxy detection that pinpoints true origin and neutralises cross-border rings.
/05
Device & OS signals
Fingerprint, OS and browser anomalies that flag emulators, spoofed setups and repeat offenders.
/+
Custom parameters
Send your own fields and aggregates, then score against them with fully tailored rules.
Core Technology
The Scoring Engine architecture.
Each transaction crosses four stages — from raw signal to final verdict — every step logged and returned for forensic certainty.
Step /01
Enrich the signal
Incoming data is supplemented with extra data points derived from the API request.
Step /02
Analyse for risk
Each data point is checked across footprint, device, IP, email and phone.
Step /03
Score the event
Rules with a score action sum their values into the fraud score.
Step /04
Decide the verdict
The score resolves to approve, review or decline — with the full rule trail.
Decision Intelligence
One signed verdict, every signal.
Atna rolls every score and module check into a single, shareable report — an overall Atna score, AI-written intel, and a per-module breakdown that separates established histories from synthetic, thin-file risk.
- Explainable scoring with full signal reasoning
- Complete rule trail in every API response
- Exportable, audit-ready verdict record
Ishaan Nair
ACZPY871OA · Verified Nov 9, 2027
Overall ATNA score
Low Risk Profile
HighMidLow
✦ Atna Score Intel
- Phone verified, normal usage, no fraud found.
- Partial match in first name — cross-record check advised.
- GST Number valid but currently inactive.
- Corporate identity verified with consistent officer details.
Digital Footprint
82%
Low RiskIntelli KYC
70%
Mid RiskIntelli 360
63%
Mid RiskTru Docs
92%
Low RiskOrchestration Workflow
How Atna scores every transaction.
One pipeline orchestrates enrichment, rules, lists and machine learning — passively, in real time, before a decision is ever returned.
Step /01
Pre-decision signal capture
Atna silently reads device fingerprint, IP reputation and behaviour to build a risk baseline with zero friction.
Step /02
Rule evaluation
Default and custom rules fire, each adding, subtracting or forcing a state.
Step /03
Risk scoring
Signals resolve into a fraud score and an explainable AI Insights score.
Step /04
Continuous monitoring
Labelled outcomes feed back, sharpening every future decision.
Transaction in
API request
Enrichment
footprint · device · IP
Rule eval
default + custom
AI & lists
insights · black/white
Verdict
approve / review / decline
Atna Engine
One engine for every decision.
Engineered to provide a multi-dimensional view of risk by combining advanced machine learning with real-time data orchestration. Rules give you immediate, explainable control; the models detect synthetic patterns legacy systems miss — together, one structured verdict.
Learn MoreDefault rulespre-built
Custom rules & parameterstunable
AI Insights score0–100
Black / white / custom listspolicy
Rule categories & hierarchygrouped
Full rule trail in API responseauditable
Key Takeaways
Built for measurable outcomes.
/01
Multi-layer decisioning
Rules, machine learning, lists and categories combine into one structured verdict — far stronger and more accurate than any single check.
/02
Real-time, explainable scoring
Every transaction is scored instantly and assigned a verdict, with the exact signals that raised and lowered the score on show.
/03
Hard policy with lists
Blacklists and whitelists set absolute outcomes; custom lists feed your rules without adding friction for trusted users.
/04
Auditable by design
Every decision ships with its rule trail in the API response and on the event timeline. Nothing is a black box.