A data source is any system that supplies the numbers a report is built from: a GA4 property, a Google Ads account, Search Console, your CRM, your shop platform or a spreadsheet. In reporting tools such as Looker Studio the term also has a narrower meaning: a saved connection to one of those systems, with its fields, settings and access rules defined once so that any chart can reuse it.
How a data source works
Every figure in a marketing report comes from somewhere, and that somewhere is its data source. Spend comes from the ad platforms, sessions and key events from GA4, organic clicks and impressions from Search Console, and enquiries or sales from the CRM or the till. Each system counts in its own way, so the same word, such as “conversions” or “users”, can mean different things depending on which source sits behind a chart.
In Looker Studio, a data source is a specific object. You choose a connector (Google’s own for GA4, Google Ads and Sheets, or a partner connector for platforms such as Meta), point it at an account, and Looker Studio lists every available field as a dimension or a metric. You can rename fields, change their types, add calculated fields and decide whose credentials the source runs on. Owner’s credentials let anyone who can open the report see the data. Viewer’s credentials show each person only what their own login allows. Several reports can then share that single data source.
GA4 uses the term as well. Under Admin, then Data import, a data source is a file or connection that adds information GA4 cannot collect by itself, such as cost data from Meta or Microsoft Ads, product details or offline sales. GA4 matches the imported rows to your existing data through a shared key, such as a campaign name or an item ID.
Why it matters
Most arguments about marketing numbers turn out to be arguments about data sources. A director sees a dashboard showing 60 leads last month, the sales team counts 41 in the CRM, and Google Ads claims 52. None of these figures is necessarily wrong. Each comes from a different source with different rules on consent, attribution and duplicates, and knowing which source sits behind each figure is what lets you decide which one to run the business on.
For a UK business there is a governance point too. A Looker Studio data source running on the owner’s credentials shows its data to everyone the report is shared with. If that source connects to a CRM or a sheet holding names, email addresses or phone numbers, sharing the report widely can expose personal data to people who have no reason to see it. That is a UK GDPR problem, not just a reporting one.
Common mistakes
- Building every data source on a freelancer’s or former employee’s personal Google account. When that account loses access, dashboards break or go blank without warning.
- Making a fresh copy of the same data source for each report, each with slightly different calculated fields, so two reports disagree and nobody can say why.
- Comparing figures from different sources as if they measured the same thing, such as GA4 key events against Google Ads conversions, without allowing for differences in attribution and counting.
- Using a Google Sheet as a source and letting colleagues rename columns or insert rows above the headers, which quietly breaks the field mapping.
- Combining two sources in a blended data source on a join key that does not match exactly, such as campaign names spelled differently in each platform.
How to act on it
List every figure your monthly report shows and write the system it comes from beside each one. That single page usually explains most of the disagreements in the room. It also shows which number should be your source of truth for each decision: spend from the ad platforms, sales from your shop or accounts, and on-site behaviour from GA4.
Next, move ownership of every data source to a shared business account rather than an individual, and reuse one data source per system across all your reports instead of copying it. Before sharing a report, check which fields each source exposes and remove anything holding personal details unless the readers genuinely need it. If you want this organised as part of paid campaign reporting, it sits within my performance marketing work, where every channel is judged against one agreed cost per lead.