The architecture: Detrics, BigQuery, master table and Claude Code

The pieces and how they connect

The dashboard’s architecture goes like this:

1. Detrics acts as the data source: it feeds BigQuery with all the data from each ad platform and from GA4.
2. BigQuery holds the raw data, and that’s where a master table gets created to consolidate every platform’s information.
3. Claude Code builds the dashboard that reads from that table.
4. Vercel hosts the deployment.

The master table is what solves the attribution problem: it blends GA4’s purchases and revenue with each platform’s spend and campaigns, matching on the Campaign ID UTM.

All the code, in a public repo

The dashboard in the video is on GitHub: github.com/detrics/dashboard-marketing-digital-bigquery. You can clone the code directly, start playing with Claude Code and build your own version on that base. And the deployed result is live at dashboard-demo-playground.vercel.app.

The key insight

Once the raw data is consolidated in BigQuery, everything is ready to play: you hand Claude Code a BigQuery with all the information, it reads the tables, understands perfectly what you’re going for, and can create the master, build the dashboard and more with no trouble. The heavy lifting is the data; the rest is agentic AI.

The step by step for the rest of the series: create the transfers with Detrics, consolidate the master in BigQuery, build the dashboard with Claude Code, and deploy on Vercel.