You observe that customers who used a new support feature at least once had a 30-day retention rate of 68% (816 of 1,200), while customers who did not use it had a 30-day retention rate of 60% (1,080 of 1,800). However, high-intent customers were more likely to use the feature: among customers with high prior engagement, retention was 80% for users and 78% for non-users; among customers with low prior engagement, retention was 50% for users and 48% for non-users.
How would you use these numbers to explain the difference between correlation and causation in customer behavior, and what conclusion would you draw about whether feature usage itself caused higher retention?