Taking Action on Verisoul Decisions

After integrating Verisoul and receiving decisions on accounts - either through the API or dashboard - you’ll need to implement strategies to block bad users while minimizing friction for legitimate users and eliminating false positives. Verisoul provides flexible options for implementing these decisions in your system.

Implementation Approaches

There are three main frameworks for implementing Verisoul decisions:

  1. Real-time Automated Blocking via API
  2. Manual Review and Offline Batch Blocking
  3. Integration with Custom Fraud Models

Real-time Automated Blocking

Verisoul aims to automate the data science work needed for accurate user decisions. While Verisoul provides the decisions, you’ll need to implement the actual blocking mechanism in your system. This is typically done alongside an authentication provider (e.g., Auth0’s blocking API).

The decision field can be tuned in two primary ways:

1. ML Model Configuration

The machine learning models can be fine-tuned for your specific ecosystem:

  • Adjust sensitivity for multi-accounting detection
  • Configure proxy/VPN detection thresholds
  • Customize device risk parameters
  • Tune bot detection sensitivity
  • Modify email risk scoring
  • Calibrate how sub-scores contribute to the overall account score

Contact the Verisoul team to modify these configuration values. They can help evaluate different scenarios or analyze disposition data based on downstream metrics like chargebacks.

2. Lists and Rules

Lists help you organize accounts and enable specific workflows for different user groups. Verisoul provides several standard lists by default:

  • Allow List: Marks accounts as Real
  • Block List: Marks accounts as Fake
  • Main Account List: Sets multiple_accounts score to 0 (useful for allowing one primary account for users with multiple accounts)

Important Notes about Lists:

  • Standard lists (allow, block, main_account) are mutually exclusive
  • Custom lists can contain overlapping accounts
  • Lists can be managed via both dashboard and API

Rules System

Rules provide real-time logic for automatically adding users to lists. For example:

IF country != US THEN add to block list

Key aspects of rules:

  • Multiple rules can run concurrently
  • Rules respect list hierarchy (won’t override allow/block/main_account status)
  • Can operate on both account and session-level data
  • Currently implemented by the Verisoul team (self-service UI coming soon)

Progressive Security with Step-up Verifications

To minimize friction for legitimate users while maintaining strong security, Verisoul supports a progressive security framework using step-up verifications. This approach allows you to:

  1. Start with lightweight, frictionless checks for all users
  2. Progressively add stronger verification requirements only when needed
  3. Reduce false positives and support workload
  4. Maintain security without compromising user experience

Managed Verification Services

Verisoul provides several managed verification services that can be triggered based on risk signals:

  • Face Match: Verify user identity through facial recognition
  • ID Check: Validate government-issued identification documents
  • Phone Risk Detection: Assess phone number risk and ownership
  • Email Risk Detection: Evaluate email address authenticity and risk

These verifications can be implemented in various ways:

  • Proactive: Required during initial signup for high-risk segments
  • Reactive: Triggered when suspicious activity is detected
  • Progressive: Added as users access more sensitive features
  • Risk-Based: Applied based on risk score thresholds

For example, you might:

  1. Allow initial access with basic email verification
  2. Request phone verification if risk signals increase
  3. Require ID verification for high-value transactions
  4. Add face match for suspicious account patterns

This layered approach helps balance security and user experience while providing clear paths for users to verify their legitimacy when needed.

Manual Review and Batch Processing

For cases where real-time blocking isn’t required or when you want additional human oversight, you can:

  • Review decisions in the Verisoul dashboard
  • Export data for offline analysis
  • Implement batch blocking processes

Custom Fraud Model Integration

For organizations with existing fraud detection systems, Verisoul’s decisions and signals can serve as additional inputs to enhance your models. This allows you to:

  • Combine Verisoul’s insights with your internal data
  • Create custom scoring models
  • Implement graduated response systems

Next Steps