The telecom operator wanted to boost revenue by promoting higher-tier plans and additional services. They had accumulated historical data, but to unlock its value, machine learning was needed. We helped the client develop ML models to improve the conversion of up-sell and cross-sell campaigns. As a result, they gained an extra $2.28M in sales. Read the full story in our case study.

Objective: Increase Sales of Plans and Services

A mobile operator wanted to improve the performance of marketing campaigns focused on two key plans. Although they had accumulated a solid base of historical data for analysis, it wasn’t being used due to a lack of tools. That’s when the client turned to a vendor with big data expertise—and our project began.

The Eastwind team provided expert support to the operator throughout the project. Together, we set out to:

• Build analytical Next Product to Buy (NPTB) ML models to predict the likelihood of plan purchases
• Integrate these ML models into the marketing campaign workflow
• Scale the approach to other operator products.

Using the client’s data, we helped them build NPTB ML models that boosted campaign conversion rates and delivered additional revenue.
 

Solution: Targeting High-Response Subscribers and Scaling the Approach

We first helped the mobile operator build NPTB models to identify subscribers most likely to switch to higher-tier plans—and encouraged them to make the move.

Here’s how the process worked:

1. Prepared historical plan data for machine learning. We analyzed the structure and quality of the operator’s raw data, standardized formats, and cleaned it (removing outliers, gaps, and duplicates). This created a solid foundation for training ML models.
2. Built a feature store. Using subscriber data on geography, spending, behavior, and interests, we selected key features, normalized and standardized them, then ranked by priority to form a feature set.
3. Developed ML models for each plan. These models identified subscribers most likely to purchase a specific plan.
4. Generated lists of target subscribers. Every month, we created lists of those most likely to respond to an offer. The operator then sent personalized promotions to this specific audience.
5. Scaled ML models to additional services. After achieving strong results with the first models, the operator launched six new products. At first, these were promoted via telemarketing to collect initial data. Once enough statistics were gathered, we built dedicated ML models for each product.
6. Continued training the ML models. The operator collected telemarketing data on plan and service sales, using it to further train the models and improve prediction accuracy.

Through this iterative approach, we continuously refined the models and, together with the client, successfully brought the project to life.
 

Results: $2.28M in Additional Revenue in One Year

The ML models boosted the conversion of marketing campaigns, driving higher sales and subscriber ARPU. As a result, the project generated an additional $2.28M in revenue over the course of a year.



Machine learning enabled the operator to:

• Improve the accuracy of audience segmentation
• Increase conversion rates in cross-sell and up-sell campaigns
• Scale the approach to new products

The Eastwind team helped the mobile operator build a full ML lifecycle—from data preparation to integrating results into marketing campaigns. This end-to-end process delivered a measurable financial impact.

Book a consultation with us, and we’ll show you how to turn your data into higher profits for your business.