Thompson Sampling Meets Uplift Modeling: A Reinforcement Learning Approach to Subscription Growth

Artificial Intelligence

Copy Link

Join to Unlock

Unlock This Lesson

min

Thompson Sampling Meets Uplift Modeling: A Reinforcement Learning Approach to Subscription Growth

publish date

Mar 12, 2025

duration

min

Difficulty

Intermediate

Beginner

Case details

You know how marketing teams spend tons of money trying to keep customers from leaving—often without knowing if their efforts even make a difference? That’s where AI-powered uplift modeling and Thompson Sampling come in. By combining these techniques, we can pinpoint the users who actually need a nudge to stay subscribed while avoiding wasted discounts and unnecessary outreach. In this talk, I’ll break down how reinforcement learning and AI-driven analytics help businesses boost retention, personalize promotions, and grow subscriptions smarter—not harder.

Share case:

About Author

Dmitrii Zolotukhin

Chief Data Officer

Dmitriy Zolotukhin is a Chief Data Officer (CDO) with over six years of experience leading data-driven strategies in streaming, finance, and e-commerce. He has previously held key positions at Tinkoff and OZON, where he played a crucial role in implementing advanced data analytics, machine learning models, and SQL-based solutions for optimizing business performance. With a strong background in data science, analytics, and PnL optimization, Dmitriy specializes in designing scalable data architectures, enhancing decision-making through predictive analytics, and improving SQL code efficiency. His expertise makes him a valuable speaker on topics related to data-driven decision-making, coding best practices, and financial data analysis.