ClickVision

MDP Clickstream Deposit Prediction & Churn Optimization

Organization: Gamezop

Key Result & Impact Benchmark

Increased deposit conversion efficiency while reducing promotional spend and churn by 62%.

Technical Overview

ClickVision is a behavioral reinforcement learning framework created for Gamezop to predict user deposit conversion from real-time clickstream event trajectories. Formulating user navigation as a Markov Decision Process (MDP), ClickVision models session events as state transitions. By computing RL transition probability matrices alongside user demographic features, the system estimates the real-time probability of a user reaching a deposit event. This allowed Gamezop to allocate targeted promotional offers exclusively to high-intent users, reducing offer wastage.

Key Technical Highlights

Modeled multi-step clickstream event sequences as Markov Decision Process (MDP) state transitions.
Reduced wasted marketing promotional offers and user churn by 62%.
Shipped production model predictions directly driving downstream UI flow modifications.

Technologies & Frameworks

Markov Decision Processes (MDP)

RL Transition Matrices

Python

Scikit-Learn

Clickstream AI