Analyzing Customer Conversion Patterns: A Survival Analysis Approach to Multi-Channel Attribution

This report explores the dynamics of customer conversion by examining the relationship between visit behaviour and conversion outcomes across various marketing channels.By condensing customer visit data into a singular representation for each customer, we capture the time intervals from their first visit to either a conversion or their last recorded visit, thereby categorizing customers as converters or non-converters.The analysis utilizes survival analysis techniques to estimate conversion probabilities over time for different marketing channels, allowing for insights into the effectiveness of each channel in driving conversions.Furthermore, the report introduces a causal inference framework to assess the impact of offline marketing interventions, specifically television advertising, on web traffic.By employing Bayesian structural time-series models, we generate counterfactual predictions to isolate the uplift attributable to marketing efforts.This comprehensive approach highlights the interplay between digital and offline marketing strategies and provides actionable insights for optimizing customer engagement and conversion.