Payment fraud prevention and detection for online merchants.


BazPay scores every order in real time before the acquirer sees it. Device, velocity and geography signals collapse into one decision. You set the thresholds, we run the engine, and 3-D Secure 2.2 steps in only when the risk earns it.

Built for regional sellers: shops, subscription software firms and professional services. Blunt blocking costs good revenue. Tuned rules do not.

Why BazPay's fraud prevention solutions sit inside the gateway

The scoring engine, the acquirer and the dispute tooling belong to the same platform. That removes the middle layer between the signals and the settlement — the part where most payment gateway fraud management falls apart.

  • Score every order before authorisation

    Signals collect at checkout and resolve into a single decision before the acquirer sees the request. Bad orders never reach the network, so scheme ratios stay clean.

  • Approvals held steady, not blanket-blocked

    Rule outcomes carry four verdicts, not two. Step-up and review keep borderline orders in play, so payments fraud prevention does not silently cost conversion.

  • SCA exemptions applied automatically

    Low-value, risk analysis, trusted-beneficiary and merchant-initiated flags are claimed where PSD2 permits. Fewer challenges, same liability shift on the ones that do fire.

  • Signals travel with the payment

    The final decision, the rules it hit and the 3-D Secure result are stamped on the charge object and re-used as dispute evidence when a chargeback lands.

What the fraud prevention software looks at

Five signal families feed the score. None of them decides alone. Together they separate a card tester from a first-time buyer without a rules war.

  • 01Device

    Device and session fingerprint

    Each checkout carries a device fingerprint, session age, browser traits and network posture. Reused devices across many cards stand out fast, which is what card testing looks like from the gateway side.

    • Fingerprint
    • Session
    • Proxy flag
    • Headless UA
  • 02Velocity

    Velocity counters

    Count attempts by card, email, IP, device or BIN over a window you pick. Card testing shows up as a burst. The burst gets blocked before your loss ratio moves.

    • Per card
    • Per IP
    • Per email
  • 03Geography

    BIN and country match

    The card issuing country is compared with billing country and IP geolocation. A mismatch raises the score. It never blocks on its own — geography is a signal, not a verdict.

    • BIN
    • Billing
    • IP
  • 04Basket

    Order shape

    Ticket size, item count, SKU mix and first-order status all feed the score. A first order at ten times your average is worth a second look, which is what a review queue is for.

    • Amount
    • New payer
    • SKU mix
  • 05History

    Your own outcomes

    Past approvals, refunds and dispute outcomes on the same payer feed back in. Good repeat buyers get a clear path through checkout. Bad-actor cohorts are marked automatically.

    • Repeat
    • Refunds
    • Disputes

Scores land beside approval data in real-time analytics, so you can see what a rule change did within minutes. See also the platform-security view on secure online payment systems and the 3-D Secure detail on secure payment services.

The order-screening flow, end to end

Eight stages sit between the checkout button and the settled payment. Each one is visible in the dashboard and each one emits a signed webhook.

  1. Collect signals

    Hosted fields and the checkout script gather device, session and network signals in the same request.

  2. Enrich

    BIN data, IP geo and your merchant reputation on the payer are attached to the request before scoring.

  3. Score

    Your rule set runs against the enriched request. Every hit is logged with its own rule ID and weight.

  4. Decide

    One of four verdicts is returned: allow, step up to 3-D Secure 2, send to review or block outright.

  5. Authenticate

    If a step-up is required, the shopper is redirected to their issuer's 3-D Secure 2.2 challenge.

  6. Authorise

    The charge posts to the acquirer with the authentication result and CAVV bound to it for issuer scoring.

  7. Stamp the record

    Final decision, rule IDs and 3-D Secure outcome are written to the payment and to the signed webhook.

  8. Feed back

    Refunds, chargebacks and manual reviews feed the payer history so tomorrow's decisions are sharper.

Four outcomes, set by you

A rule ends in one of four outcomes: allow, step up, review or block. Thresholds are yours to move. Edits apply to the next order, with no release and no ticket.

The table shows example rules, not defaults you are stuck with. Start with ours. Tune them once you have your own order history and dispute reason-code data.

Example fraud rules and their outcomes
TriggerOutcomeType
Velocity — 4+ cards on one device in 10 minutes Block card testing
Issuing country differs from billing country Score + soft signal
Order value above your set ceiling Force 3-D Secure 2 step-up
Payer has a settled dispute on file Review queue manual
IP flagged as data centre or open proxy Score + network
Repeat payer, three clean settled orders Allow trusted list
SCA exemption available (risk analysis, low value, MIT) Frictionless Open-finance regulation

Fraud prevention tools built into the payment gateway

Every capability below is on the standard integration. Turn features on with a request flag or a dashboard toggle. No enterprise-tier gate on the primitives.

  • Rule builder

    Compose triggers on 30+ enriched fields. Publish rule changes from the dashboard — no engineering release required.

  • Velocity counters

    Windowed counts on card, IP, email, device and BIN. Card-testing bursts and BIN-attack sweeps are caught inline.

  • Device fingerprinting

    Stable device hash tracks a shopper across sessions without cookies, so trusted repeat buyers keep frictionless paths.

  • Allow and deny lists

    Curate lists of payer IDs, emails, BINs, IP ranges and shipping addresses. Bulk import from CSV.

  • Review queue

    Borderline orders park in a queue with the full signal breakdown. Analysts approve, deny or refund with one click.

  • 3-D Secure 2.2 engine

    Frictionless flow first, challenge only when your rules or the issuer require it. Exemption logic runs automatically.

  • Dispute defence

    Early scheme alerts, evidence templates and one-click submission. The 3-D Secure proof is attached automatically.

  • Signed webhooks

    HMAC-signed, replay-protected events for every decision, step-up outcome and dispute state change.

Read the risk verdict from the API and webhook

The charge response and the signed webhook carry the risk verdict, the rules that fired and the 3-D Secure outcome. Wire that into your order router, your dunning system or your CRM — the payment gateway fraud detection surface is a first-class part of the API.

{
  "id": "ch_02H9F...",
  "status": "requires_action",
  "amount": 12900,
  "currency": "EUR",
  "risk": {
    "verdict": "step_up",
    "score": 46,
    "rules_fired": ["velocity.ip.5min", "geo.mismatch"],
    "sca": { "type": "3ds2", "exemption": null }
  },
  "next_action": { "type": "3ds_redirect", "url": "..." }
}

Every state change fires a signed webhook — charge.reviewed, charge.blocked, dispute.opened. The complete event catalogue lives in the API reference and the developer docs.

3-D Secure 2.2 without the drop-off

Strong customer authentication is an PSD2 duty. It is also a common place to lose sales, because every prompt adds friction. We ask for one only when the rules say the risk earns it — and when we do, the issuer already has the enriched risk data, so most challenges resolve frictionlessly.

Low-value, risk analysis, trusted-beneficiary and merchant-initiated exemptions are requested where PSD2 allows. When a challenge does fire, liability for that order shifts to the issuer. The result is stored with the payment and re-used as dispute evidence.

Cards are one rail among several. Open banking carries bank-side authentication of its own, with no card number in the flow at all — a useful addition to your online payment fraud prevention posture for higher-ticket orders.

More on the rails: card and APM processing and recurring billing for saved-card renewals.

Fraud prevention online payment scenarios we tune for

The default rule set already reflects these four merchant profiles. Turn the ones you need on from the dashboard and adjust the thresholds against your own order history.

  • E-commerce sellers

    Card-testing burst detection, BIN attack blocking and address-mismatch scoring for DTC and multi-country storefronts.

  • Subscription software

    Merchant-initiated exemptions on renewals, dunning-aware velocity rules and dispute reason-code webhooks.

  • Professional services

    Higher-ticket B2B invoicing with named-payer allow lists and enforced 3-D Secure 2 over your chosen ceiling.

  • Digital publishers

    Account-takeover heuristics on membership renewals and shared-credential detection across paywall unlocks.

Out of scope for BazPay: adult, gambling, CBD, nutraceutical, forex, CFD, crypto-exchange, debt-collection and MLM.

Inline rule engine vs manual review

Digital payment fraud prevention only works if it runs on every charge, at the moment of the charge. The comparison below explains why a batch-review process leaves losses on the table that an inline decision would not.

BazPay inline rule engine compared with manual batch review
Dimension BazPay inline engine Batch review workflow
Decision latency Inline, sub-second Batch review, hours
Coverage Every charge Sampled or above-threshold only
Verdicts Allow / step-up / review / block Allow / hold
Rule changes Dashboard, effective on next order Code deploy or PSP ticket
3-D Secure link Fires exactly when the rule earns it Blanket or off
Dispute evidence 3-D Secure + signals auto-attached Manual packet assembly

Fees and settlement details on the pricing page. Plugin coverage on the integrations page. For the outbound side of the flow, see payouts. The rule engine sits inside the wider payments solutions bundle, boarded under merchant acquiring, and runs on the same payment infrastructure — with dispute defence tuned for intangibles on the digital-goods gateway.

When a dispute does land

Screening cuts the volume. It never gets it to zero. So the case flow is built in.

  1. Early warning lands

    Schemes send pre-dispute alerts (RDR, Ethoca, CDRN where enrolled). We surface them in the dashboard and by signed webhook.

  2. Refund or defend

    Refund early and the case often stops before a chargeback opens. Choose to defend and we open a slot for your evidence.

  3. Attach evidence

    Upload delivery proof, session logs and the 3-D Secure 2 result. The packet is formatted to the scheme's current template.

  4. Track the outcome

    Every state change is stamped in the dashboard. Export the case list as CSV for your finance team's reconciliation.

Common questions

How does BazPay's payment fraud prevention differ from a standalone fraud prevention tool?

The rule engine, 3-D Secure 2.2 and dispute tooling ship inside the same payment gateway that clears the transaction. That removes a hop and gives the score every enriched field the acquirer has — BIN metadata, network-token status, issuer response history — that a bolt-on fraud prevention solution cannot see.

Do I have to write my own rules?

No. Sensible defaults for card testing, velocity, geo mismatch and step-up ceilings run from day one. You can then tune each threshold from the dashboard, with no code change and no release. Rule versions are logged, so you can revert a change if it moves the wrong metric.

Will fraud screening slow my checkout?

Scoring runs inline, before the authorisation call. Shoppers see one flow. Only step-up verdicts trigger an additional issuer prompt through 3-D Secure 2 — which for most orders is frictionless because the risk data has already been shared with the issuer.

Does the payment gateway fraud detection see raw card data?

No PAN reaches your servers. Hosted fields and gateway-side vaulting keep card data inside our PCI DSS Level 1 scope, so your annual return stays at SAQ A. The scoring engine works on tokens and enriched signals, never on the card number in the clear.

How is dispute evidence prepared?

Every settled order carries its 3-D Secure 2 result, the signals that fed the decision and the delivery log if you push shipping events to the API. When a chargeback opens, the evidence packet is pre-assembled to the scheme's current template — you review, add anything specific and submit.

How does the system catch card testing and BIN attacks?

Velocity counters run on card, IP, email, device and BIN over configurable windows. A burst of attempts against sequential card numbers on one device triggers an automatic block, and the source device is added to a rolling deny list without human intervention.

Do you handle online payment fraud on recurring subscriptions?

Yes. Merchant-initiated transaction exemptions run on renewals with a stored mandate, so trusted repeat charges stay frictionless. Dunning-aware velocity rules distinguish a genuine card-update flow from an account-takeover attempt.

Which merchant profiles is this suitable for?

BazPay is a regional acquirer. Merchant fraud prevention here is tuned for e-commerce sellers, subscription software firms, professional services and digital publishers. BazPay does not board adult, gambling, CBD, nutraceutical, forex, CFD, crypto-exchange, debt-collection or MLM merchants.

Test your rules on sandbox data

Replay approvals, declines, step-ups and disputed charges before a single live order. Sandbox keys are self-service and the hosted checkout products are wired in by default.