Context

StripeCart uses a rule-based fraud screening layer before payment authorization. One high-severity rule automatically declines card transactions when the billing country and IP country differ and the account is less than 7 days old. Over the last 6 weeks, customer support complaints about “card declined” events increased, and the Risk team wants to know whether this rule is blocking too many legitimate purchases.

Current Performance

The Problem

The rule catches fraud, but most declined transactions appear to be legitimate. You need to determine whether the rule is too aggressive, how to quantify the business tradeoff, and what changes should be made.

Requirements

1. Interpret whether the current decline rate is acceptable using the metrics above.
2. Explain which additional analyses you would run to verify that the rule is causing excessive legitimate declines.
3. Identify likely failure segments and how you would validate them.
4. Recommend rule or threshold changes that reduce customer friction without materially increasing fraud loss.

Constraints

1. Fraud losses above $600K/week are unacceptable.
2. Payments leadership will not accept more than a 0.5 point drop in approval rate.
3. Manual review capacity is limited to 4,000 transactions per week.