Contact

Tools · Measurement & data

BigQuery

One set of numbers under every report.

BigQuery is Google Cloud's serverless data warehouse: you store large tables and query them with SQL. Donut Analytics keeps each client's ad, analytics and sales data there, rebuilt on a schedule, so dashboards, budget pacing and media mix models all read the same numbers. Where a platform has no connector, we write the pipeline in Python.

Let’s Chat All tools

How we use it

How we use BigQuery

  • Daily loads

    Ad platform data lands in BigQuery every day through a connector service, and scheduled SQL rebuilds each client's media tables, daily or hourly.

  • One media table per client

    We write SQL that puts every channel, including TV and streaming, into one table with the same columns, so you compare channels on the same terms.

  • Pipelines where there's no connector

    When a platform has no connector, we write one in Python. Our pipelines load partner APIs, media plans, TV plans and even emailed sales reports, and they flag it when data goes stale.

  • Checked against the source

    We check warehouse totals against the ad platforms themselves. When a feed goes wrong, we correct it in the warehouse and say so.

  • Conversions that arrive late

    Google's store visits keep arriving for weeks after the ad. We estimate recent days from each campaign's own history and take daily snapshots to measure how fast the real numbers fill in.

  • Daily budget pacing

    Spend is paced every day against the monthly or flight budget, straight from the media tables, so overspend shows up early.

In practice

What's in the warehouse

Real work, kept anonymous.

  1. 01

    The data behind a media mix model

    For a Florida retailer, BigQuery holds the weekly sales and media data behind the media mix model we refresh every month, plus the outside factors it controls for, like search interest in the category.

    Read more
  2. 02

    TV plans next to digital

    For one client, TV plans arrive as spreadsheets. A Python script turns them into estimated daily spend, so TV sits next to digital in the same table.

  3. 03

    Sales history from the client's own reports

    We built one client's daily sales history from old exports plus a report that arrives by email each day, checked against their own totals, so ad results can be read against real sales.

  4. 04

    Store visits, discounted

    We don't take platform store-visit counts at face value. We discount them with rates checked against the media mix model, so a dashboard's return on ad spend tells the same story as the model.

What it can’t tell you

What BigQuery can't tell you

  • It stores what you give it

    BigQuery doesn't collect anything on its own. If a platform's export duplicates rows or skips a day, the warehouse holds the mistake until someone checks it against the source.

  • Comparable isn't causal

    A warehouse makes platforms comparable. It can't say which of a platform's claimed sales your ads really caused. That takes a lift test or a media mix model.

  • Costs grow with use

    Past a free monthly tier, Google bills for storage and, on standard pricing, for the data each query scans. A careless query on raw event data can cost real money.

Proof

From our results

Real client work, kept anonymous. Each number links to how it was measured.

FAQ

Questions about BigQuery

Do I need BigQuery if I already have GA4?
Not always. If GA4 and one ad platform answer your questions, you may not. Once you want every channel's spend next to your sales, longer history, or a media mix model, a warehouse is the simplest place to keep it.
How much does BigQuery cost?
There's a free monthly tier of 10 GiB of storage and 1 TiB of queries, and a sandbox you can try without a credit card. Past that, Google bills for storage and for the data each query scans. Prices change, so check Google's pricing page.
Can Google Ads and Meta data go into BigQuery?
Yes. Google's own transfer service has connectors for both, and connector services do the same. Donut Analytics uses a connector service for the major platforms and writes Python pipelines for the rest.
Can GA4 export to BigQuery?
Yes. Standard GA4 properties can export raw events to BigQuery. The daily export is capped at 1 million events a day. The streaming export has no cap but adds BigQuery costs.
Do I need a data engineer to use BigQuery?
Someone has to write and maintain the SQL and the pipelines. When Donut Analytics runs it for a client, that's our job. Your team gets the dashboards and models it feeds.

Pairs with

Guides: How to measure marketing when most sales happen in stores FAQ

All tools

Put every channel in one place.

Tell us what you’re working on. We’ll build a measurement-first plan for your brand. Email hello@donutanalytics.com or call (407) 394-9720.

Let’s Chat See our results