Services · Marketing data warehouse
All your marketing data in one environment you own.
Your marketing data sits scattered across platform exports, spreadsheets and separate dashboards. I set up one environment where ad spend, on-site behaviour and revenue live in the same tables, with fixed definitions and a daily refresh.
01 / Problem
Recurring problems
Where it usually goes wrong
Three things I run into on almost every assignment.
01 · Ownership
Your data is locked inside the tool
If the connector hangs straight off your dashboard, the data never lands anywhere that belongs to you. Cancel the subscription and the history goes with it. You pay monthly for access to your own figures, and outside that one dashboard you can do nothing with them.
02 · Analysis
Numbers without an explanation
A chart shows revenue falling, but not which campaign or which segment caused it. That calls for a data model where cost, behaviour and revenue are linked to each other, not for another chart.
03 · Data quality
The dashboard doesn't match the ad platform
Standard connectors only load the new day, while platforms assign conversions retroactively. Someone clicks on Monday and buys on Thursday: the platform writes that conversion back to Monday, but in your warehouse Monday is already fixed. The gap grows every week.
Standard connector
42
conversions · new day only
Google Ads interface
50
conversions · assigned retroactively
The fix isn't complicated: reload the last 30 days every night instead of just yesterday. It costs a little more compute and saves a lot of argument on Monday morning.
02 / Setup
From source to reporting
How it fits together
You don't need to know up front what that data model should look like. We start with the questions you currently can't get an answer to and work back to the sources you need for them.
-
Fivetran
Source
Managed connectors to Google Ads, Meta, GA4, your shop and your back end. No home-grown scripts that fall over on an API change.
-
BigQuery
Storage
One warehouse in your own Google Cloud project. You own the data and the invoice.
-
dbt
Transformation
All logic as SQL in version control, with tests on the output. A KPI is defined in exactly one place.
-
Looker Studio / Power BI
Reporting
Dashboards on top of the model, so every number traces back to its source table.
Curious what such a dashboard looks like in practice? See the live dashboard demo.
03 / Result
Delivery and costs
What you get and what it costs
What's there when it's done
- A BigQuery project in your own company's name, with all your sources in it.
- A data model in dbt where every KPI is defined once.
- A dashboard in Looker Studio or Power BI, refreshed every morning.
- Documentation and access, so somebody else can take it over.
- All accounts and subscriptions in your company's name, not mine.
- If you stop, everything loaded up to that point simply stays.
What it costs per month
Everything is in your own company's name, so you see exactly where the money goes. No subscription runs through me.
- BigQuery
- Usage-based. Marketing data is small compared to what BigQuery handles.
- Ads, GA4
- Free export. You only pay the BigQuery side.
- Other
- Meta, LinkedIn, shop or CRM through Fivetran, in your name.
- Reporting
- Looker Studio is free. Power BI charges per user.
- Groen Data
- Only during the build, nothing after.