Project details:
Analyzed real mobile usage data to compare two legacy tariff plans and determine which one brings higher monthly revenue. The project combines data cleaning, exploratory analysis, and statistical testing to identify the more profitable plan and support a data-driven marketing strategy for the operator.
Description
Business Context & Problem
Many customers of a mobile operator continue using outdated tariff plans, and the marketing team needs clarity on which of the two available tariffs actually generates higher revenue. My task was to turn raw usage logs into a clear comparison and show which plan deserves more advertising support based on real customer behaviour.
Data & Analytical Approach
I worked with anonymised telecom data that included monthly records of calls, SMS, and internet traffic by user and tariff. After cleaning and preparing the dataset, I calculated monthly revenue per customer and reconstructed how their real usage aligned with the limits of each plan. This allowed me to explore usage trends, segment behaviour by region, and identify where the revenue differences originate.
Statistical Testing & Validation
To confirm the findings, I approached the tariff comparison as an A/B-style test. I applied descriptive statistics and conducted hypothesis testing to evaluate whether the revenue difference between the two tariffs is statistically significant. A second hypothesis focused on geographic behaviour—whether Moscow users generate more revenue than users in other regions under the same plans.
Key Insights & Final Recommendations
The analysis showed a clear and statistically supported revenue difference between the two tariffs, allowing me to confidently identify the more profitable one. The regional comparison also challenged assumptions about Moscow versus non-Moscow users, providing valuable context for targeted marketing. As a result, the operator receives a grounded, data-driven recommendation for how to prioritise its marketing budget and which tariff to promote moving forward.