Marketing BigQuery Dashboard

Rebuild Relief run lead generation across Google and Meta for hail and storm damage claims throughout Queensland. Spend decisions were being made inside each platform's own reporting, which meant nobody could see what a lead actually cost once both channels were counted together.

The pipeline pulls from the Google Ads API into BigQuery, transforms it in Python on a nightly schedule, and surfaces cost-per-acquisition in Looker Studio alongside custom charts. Credentials sit in Secret Manager rather than in the job itself, and the warehouse is structured as dimension tables around a daily metrics fact table so a question about one campaign does not require reprocessing the lot.

It now carries reporting for the wider tool estate as well — around 50,000 data points a day across the applications built for that business.

How it came together.

The work moves from the right question to a system the team can actually use.

1. 01

   Strategy

   It starts with a strategy session — I come to you, learn the business inside out, and shape a plan you actually want and that actually works. Goals first, never template-first.

2. 02

   Design

   Then I design it — interface, brand and the feel of using it — mapped straight to that strategy. One person, one vision, no handoff gap between designer and developer.

3. 03

   Development

   I build the real thing: fast, typed end-to-end, Supabase-backed, and yours outright — the website and the operations behind it, shipped on a tight loop.

4. 04

   Iteration

   We iterate on the live product — real usage, real numbers — tightening the funnel and the workflows until it genuinely performs.

5. 05

   Ongoing management

   From there I keep it running and growing — performance, updates, new campaigns and features as the business moves. One number to call, no agency runaround.

Role

Data engineering

Year

2025

Client

Rebuild Relief