Context

StreamWave uses a binary classification model to predict whether a paid subscriber will churn in the next 30 days so the CRM team can send a retention offer. The current model ranks users reasonably well, but the retention team believes the operating threshold is too conservative and is missing too many at-risk customers.

Current Performance

Validation set size: 100,000 subscribers. Actual 30-day churn rate: 12%.

The Problem

The VP of Growth wants a threshold recommendation for next month’s campaign. You need to evaluate the false positive vs. false negative tradeoff and determine whether the current threshold should be changed.

Requirements

1. Interpret the business impact of false positives and false negatives at both thresholds.
2. Recommend which threshold StreamWave should use and justify it quantitatively.
3. Explain whether precision, recall, or another metric should drive threshold selection here.
4. Describe what additional analyses you would run before finalizing the threshold.
5. Suggest model or process improvements to reduce the tradeoff over time.

Constraints

• Monthly outreach capacity: 20,000 users
• Offer cost: $8 per contacted user
• Saved margin from prevented churn: $70 per true positive
• Threshold must be explainable to marketing and finance stakeholders