← Back to Projects

TV Ads Attribution 📺

Project Details

This tool measures the short-term online impact of TV advertisements by analyzing post-spot visit spikes and calculating website traffic attribution. This system helps businesses evaluate the effectiveness of their TV campaigns by tracking key performance indicators (KPIs) such as website visits and conversions. It supports TV spots from various sources and integrates seamlessly with user tracking platforms like Google Analytics, Mapp, and Adjust, enabling a comprehensive evaluation of TV campaign performance across multiple channels.

TV Ads Attribution Visualization

The system hypothesizes that within 8 minutes of a TV spot airing, the business partner’s webpage will see a surge in visits. Ideally, traffic spikes immediately after the ad, then gradually declines, returning to pre-ad levels by the 8-minute mark. All visits exceeding the baseline during this window are attributed to the TV spot, assuming they were influenced by it.

Key Features

My Contribution

I began as an Associate, assisting with the refactoring and cloud migration of an on-premises solution. I quickly took ownership of the product, leading feature development and enhancements. Notably, I designed an AI-driven attribution algorithm that significantly improved accuracy. I also strengthened the codebase by implementing comprehensive test coverage, a command-line interface for streamlined reprocessing, and various optimizations to enhance performance, scalability, and reliability. After four years leading the project, I am recognized as the company's expert in this field.

Challenges and Solutions

Outcome

Technologies

Python ETL Data Engineering PySpark AWS Glue AWS DynamoDB AWS Athena AWS Lambda AWS EC2 AWS S3 AWS Kinesis AWS Step Functions AWS CloudWatch AWS SNS Google BigQuery Google Analytics Mapp Adjust Nielsen