What a finished ad data warehouse looks like in BigQuery

The ad-level TikTok table

The video opens with the destination: in BigQuery, an ad-level TikTok table with all the relevant dimensions (campaign, ad group, ad), each ad’s image URL and all the metrics. That table holds almost 300 thousand rows with data from 30 different ad accounts.

The most important detail is in the image column: the creative is hosted in your own bucket. If you’ve worked with ad-level data before, you know the API returns a URL that expires after 7 days and breaks any report. Not here.

The Meta table

Meta has an equivalent table: dimensions at campaign, ad set and ad level, plus all the metrics. Almost 3 gigabytes of data and 3 million rows. Strings stored as strings, numbers as numbers, dates as dates: every data format Meta uses comes out properly typed. And the image, just like TikTok, in your own bucket.

The unified view: Meta + TikTok in one table

Finally, a BigQuery view unifies both platforms’ data in a single table. Each platform names things its own way; this view brings everything to a common naming for campaign, ad group, optimization goals, ad and metrics.

That makes consumption much easier: you connect the unified view to a report, an alerting system or whatever you want, without querying two separate tables. This query was built with Claude using the BigQuery MCP, and later in the series you’ll see exactly how.

With this, data from 60 ad accounts lives unified in one place. And it’s extensible: you can add tables for every platform you work with and keep folding them into the unified view.