2025·Data Analyst

Custom Attribution Model on BigQuery

Built a data-driven, position-weighted attribution model on top of the raw GA4 BigQuery export, replacing last-click reporting.

BigQuerySQLdbtLooker StudioGA4

The marketing team's last-click GA4 reports systematically under-credited upper-funnel channels. Designed a SQL-based multi-touch attribution model over the raw BigQuery export — flattening nested event/session data, reconstructing user paths, and weighting touchpoints — surfaced through a Looker Studio dashboard refreshed nightly.

Highlights

Reduced query cost 90% by materializing flattened event tables
Reallocated ~15% of paid budget toward upper-funnel channels based on model output