Select all lending products currently active. We use your product mix to evaluate your regulatory exposure and data requirements — different products carry meaningfully different fair lending and model governance obligations.

Count of credit applications received monthly, across all products. We use this to assess whether your decisioning approach is appropriately scaled to your application load.

Select the approach that best describes your primary credit decisioning method today. If you use multiple approaches, select the most advanced one currently making live credit decisions.

When was your credit decisioning model or ruleset last formally validated for performance drift, threshold calibration, and regulatory alignment? SR 11-7 requires ongoing model validation — this obligation sits with your institution, not your vendor.

CFPB examiners now require institutions using AI/ML models to demonstrate both disparate impact testing and an active search for less discriminatory alternatives (LDA). Vendor-managed testing does not satisfy this obligation if your institution cannot produce the underlying methodology and results.

ECOA and Regulation B require that adverse action notices reflect the specific reasons the applicant was denied. For AI/ML models, generic reason codes do not satisfy this requirement. CFPB has issued enforcement actions on exactly this issue.

When your risk team identifies a need to update an underwriting rule — tightening a cutoff, adding a new condition — how long does it take to deploy that change? Engineering dependency is the primary driver of decisioning lag at active lending programs.

Synthetic identity fraud and bust-out fraud typically enter through the application stage. Without a dedicated fraud layer, these losses are attributed to credit defaults — corrupting your model training data and understating your true credit risk.

Select the priority that best reflects where your credit risk program is focused right now. This shapes the recommendations in your assessment.