New features now available on Fraud Detection

We've recently released several features on our Fraud Detection product to help you continue keeping safe from online payments fraud.

Back testing (coming soon)

If you use our Fraud Detection Pro solution, you now test risk strategies using historic transaction data and iterate on rules and outcomes at a much faster rate. To perform a backtest, navigate to the risk strategy page of the tool, make a change to a strategy and press ‘save and begin test’.

You will then see an option to perform a backtest. Select the date range against which you would like to test the transaction, and results will be available in a matter of minutes:

You will be able to see a breakdown of your current live strategy performance against your test strategy. This allows you to compare outcome distributions (e.g. split of declined vs 3DS vs accepted traffic) and blocked fraud:

Device ID velocities and data quality widget

You can now create velocity rules which leverage device fingerprint. You can find the property in both the simple and relative velocity section of the rule builder, and can use this to check how many cards are being used per device:

This is available if you use a Checkout.com hosted form to process payments (Frames, Hosted payment pages and Payment links) with no additional integration required. If you use our full card details API, you'll need to integrate with our risk.js package to send us device data.

Integrating with this package also enhances our Machine Learning model performance.

You can see what other data points you are providing and their importance by clicking the Risk tab of your dashboard, and scrolling to the bottom of the page. We strongly recommend you view this table and pass us as many data points as possible:

Decline rules performance

View at a high level which rules are declining the most transactions over certain periods of time, and the trend in their decline rates. This is useful for analyzing where you may want to make changes to your risk strategy:

New properties to build rules

Fraud Detection Pro users can now use any of these attributes to build their rules:

Pre-AUTH properties

• :CardCategory: - commercial or consumer
• :IsMerchantInitiated: - True if MIT (merchant initiated transaction), False if CIT (customer initiated transaction)
• :payment_type: - MOTO, Recurring, Regular, Installment
• :token_type: - googlepay, applepay

Post-AUTH properties

• :cvv_check_passed: - whether CVV check was successful, i.e. = Y
• :has_liability_shift: - whether payment has liability shift, determined from ECI values
• :is_3ds_authenticated: - whether the payment went through 3D Secure authentication
• :liability_shift_eci_value: - actual ECI value returned for better precision
• :cvv_code: - actual CVV code returned for better precision
• :flow_type: - whether a 3DS payment was authenticated frictionlessly or challenged with 2FA. Can be ‘frictionless’ or ‘challenged’.