Skip to content
Radar

Fraud decisions, made inside the payment.

Radar scores every transaction before it is sent for authorisation, applies the rules your business actually needs, and sends the grey area to a person instead of turning a customer away.

Scoring · Rules · Review queue · Identity signals

Risk analysisLive
$1,299.00Card ···· 4242
Risk score12/100 · Low risk
  • Device and location match previous orders
  • Card verified with the issuer
  • No custom rule triggered
Approvedsettled
Every transaction is scored before approval: a low-risk payment is approved automatically, a high-risk one is held for manual review.
Scored before authorisation
Device, location, velocity, card behaviour and the customer's history with you, weighed as the checkout completes.
Rules you write
Your business knows what a suspicious order looks like better than a generic default does. Change any rule without a deploy.
A queue, not a wall
Borderline orders wait for a human. A real customer waits minutes instead of being told no.
Identity when you need it
Escalate to document checks, facial verification and liveness on the orders that warrant it.
How it works

From signal to decision, in the time it takes to press pay.

The analysis happens inside the payment, not as a batch someone reviews the next morning.

  1. Signals are collected

    Device fingerprint, network location, order shape, velocity and the customer's own track record on your account.

  2. Your rules run

    The score meets the thresholds you set. A rule can approve, send to review, escalate to identity checks, or block.

  3. A decision, or a second look

    Clear cases resolve instantly. The rest land in a review queue with the evidence already attached.

Your rules4 active
  • Velocity

    more than 3 attempts in 10 minutes

    Review
  • Order value

    above your high-ticket threshold

    Review
  • Geo mismatch

    card country different from IP country

    Block
  • Trusted customer

    three settled orders in the last year

    Approve
A list of fraud rules the merchant defines, each mapping a condition to an approve, review or block action.
What it reads

The signals behind the score.

None of these decides anything alone. It is the combination, measured against how your own account normally sells.

Device and browser

Whether this machine has been seen before, and whether it looks like the one that bought last time.

Location

How the network location lines up with the card, the billing details and the delivery country.

Velocity

How many attempts, how fast, and how many cards have been tried against the same order.

Card behaviour

Issuer responses and patterns that separate a typo from a card being tested.

Customer history

A returning customer with settled orders is not the same risk as a first-time one.

Identity escalation

Document checks, facial match and liveness, triggered by a rule rather than applied to everyone.

Disputes

When one gets through anyway.

No system catches everything. What matters then is how fast you can answer, and whether the evidence is already where you need it.

  • Every dispute opens next to the payment, with the transaction record attached
  • Deadlines surfaced before they pass, not after
  • Evidence uploaded once, from the same screen
  • Outcomes reported back into your dashboard and your webhooks
FAQ

What teams ask first.

Will Radar block customers who were going to pay?

That is the failure mode it is designed against. Anything the model is unsure about goes to a review queue rather than a decline, so a real customer waits minutes instead of being turned away.

Can I change the rules myself?

Yes. Rules are yours to write, adjust and switch off from the dashboard, and each maps a condition to approve, review, escalate or block.

Does it work across payment methods?

Risk analysis runs across the methods on your checkout. The signals that matter differ by method — an on-chain payment cannot be charged back, for instance — and the rules account for that.

How does identity verification fit in?

As an escalation, not a toll booth. A rule can require a document check, a facial match or liveness on the orders that justify the friction, and leave everyone else alone.

Do I need a fraud analyst on staff?

No. The defaults are meant to be usable on day one, and the review queue is designed so someone on your support team can work it without specialist training.

Keep the fraud out and the customers in.