At Melligold, I measured the impact of lowering the minimum trade amount using Interrupted Time Series, contributing to a 25% lift in first-purchase conversion. I also used Propensity Score Matching to show how slow withdrawals raised churn risk, with the highest risk among VIP and whale users.
At Melligold, withdrawal fixes drove a 30% rise in NPS, while my nine-dimension behavioral segmentation surfaced a strategic whale segment. That work reshaped OKRs for two quarters and cut whale churn by 35%.
At Yektanet (Jaryan Social), I built a campaign recommendation algorithm that cut recommendation-stage effort by 50% and raised customer acceptance by 20%. I also redesigned CPM campaign logic and tested a media-plan sorting algorithm, improving CPM profit, conversion, and average order value.
My work has also included market research, product analytics, and transportation modeling. At Oxford Brookes University, I ran seven surveys and multivariate analyses of consumer behavior in UK fashion and cosmetics; at AvATS, I built choice and transformer-based models of travel-behavior drivers.
