Marketing Attribution and ROI Analysis Pipeline

This project implements an end-to-end multi-channel marketing attribution and ROI platform. It ingests touchpoint data from six source systems (Google Ads, Meta Ads, TikTok Ads, email marketing, web clickstream, CRM conversions), unifies it via identity stitching into a single user-journey graph in Snowflake, and applies five different attribution models in parallel: First-Touch, Last-Touch, Linear, Time-Decay, and a removal-effect Markov-chain model implemented as a dbt Python model executing in Snowpark.