Case study
Marketing Data Pipeline (ELT)
Automated ELT pipeline pulling ad spend from 4 platforms into BigQuery, transformed with dbt and orchestrated on Airflow — one trusted marketing source of truth, refreshed daily.
How do you turn messy marketing data (elt) sources into one dependable feed?
Using Python, Airflow, BigQuery, dbt, and SQL, the data was ingested, cleaned, and structured — automated ELT pipeline pulling ad spend from 4 platforms into BigQuery, transformed with dbt and orchestrated on Airflow…
Automated ELT pipeline pulling ad spend from 4 platforms into BigQuery, transformed with dbt and orchestrated on Airflow — one trusted marketing source of truth, refreshed daily.
Hover a row to edit · changes save to your portfolio
Problem
Marketing spend lived in 4 dashboards that never agreed on ROAS.
Approach
- Python extractors for Google, Meta, TikTok, and LinkedIn Ads APIs
- Landed raw JSON in BigQuery, modeled staging → marts in dbt
- Scheduled and monitored the DAG in Airflow with freshness tests
Result
A single daily-refreshed ROAS mart replaced 4 conflicting reports and cut the monthly reporting scramble from ~6 hours to zero.