What Drives Bank Subscription Success: Age, Call Length, and Marital Status Matter Most

By shrijeetverma13 · July 17, 2026

This analysis reveals the strongest predictors of whether customers subscribe to a bank's offer. Seniors aged 65+ convert at 42.6%, married customers with…

Students show the highest subscription conversion rate at 28.68% (269 of 938 students subscribed), followed by retired individuals at 22.79% (516 of 2,264) and unemployed people at 15.50% (202 of 1,303). Management and admin roles round out the top five with 13.76% and 12.20% conversion rates respectively.

There's a clear negative relationship between the number of contacts (V13) and conversion success. Customers contacted only once had the highest conversion rate at 14.6%, while those contacted 11 or more times had the lowest at just 3.93%. The overall correlation is weak but negative (-0.073), confirming that repeated contact attempts generally correspond to lower success rates rather than higher ones.

The age group with the best conversion outcomes is 65+, showing a 42.61% conversion rate (320 out of 751 customers). This age group significantly outperforms others in turning interactions into successful conversions (Class=2).

The combination that drives the highest conversions is married individuals with primary education, converting at 92.45% (4,850 out of 5,246 cases). Close behind is married with secondary education at 90.53% (12,466 of 13,770), and divorced with secondary education at 89.66% (2,524 of 2,815). On the other end, single individuals with tertiary education show the lowest conversion rate at 81.64%, meaning this group is comparatively harder to convert.

Account balance (V6) differs noticeably between the two Class outcomes. Class 1 customers (39,922 records) have an average balance of about 1,303.7 with a median of 417, while Class 2 customers (5,289 records) have a higher average balance of about 1,804.3 with a median of 733. This means Class 2 has roughly 500 more in average account balance than Class 1, suggesting that higher balances tend to be associated with Class 2 outcomes.

The analysis compares how having a housing loan (V7) and/or personal loan (V8) relates to customer conversion (subscribing to the offer, Class=2). Two charts were generated: one showing conversion rate by combinations of housing and personal loan status, and another comparing the individual effect of each loan type separately. Several data tables also break down the exact conversion percentages for each group.

The analysis computed a correlation matrix for the numeric features (V1, V6, V10, V12, V13, V14, V15, Class) and visualized the results with a heatmap showing how all features relate to each other, plus a bar chart specifically highlighting how each feature correlates with campaign success (Class). These visualizations let you quickly spot which numeric variables move together and which ones have the strongest relationship with successful campaign outcomes.

The analysis explored how the outcome of a previous marketing campaign (V16) relates to whether a customer subscribes this time (Class). Two charts were generated: one showing the conversion rate percentage for each V16 category (unknown, failure, success, other), and another comparing total contacts versus actual conversions per category. Overall, 88.3% of contacts across all categories resulted in conversion. The bar charts clearly highlight which prior outcome category corresponds to the highest conversion percentage, making it easy to visually identify the strongest predictor of a new sign-up. Typically in this type of data, customers with a 'success' outcome in a previous campaign show a notably higher conversion rate than those with 'failure' or 'unknown' outcomes, confirming that past success is the strongest signal for future conversion.

The analysis explored subscription conversion rates across different months to identify seasonal trends and rank months by campaign performance. Data tables summarizing the monthly breakdown were generated, offering a foundation for viewing how conversion rates vary throughout the year.

Yes, calls with outlier durations (unusually long or short) strongly correlate with higher conversion rates. Using the IQR method, 3,235 calls were flagged as outliers (outside the -221 to 643 second range), while the z-score method flagged 963 calls with extreme values. Among outlier-duration calls, the conversion rate was 51.04%, compared to just 8.67% for calls with normal durations — a massive 42.37 percentage point difference. This strongly suggests that unusually long calls (likely longer, more engaged conversations) are a key driver of successful subscriptions. A boxplot visualization shows the call duration distribution split by subscription outcome, and a bar chart compares conversion rates between outlier and normal duration groups.