How a data source works
Adding a data source means authorising a reporting tool to read from a platform and choosing exactly what it may read: which ad account, which analytics property, which website property. From then on the tool requests rows through that platform’s interface for developers, and those rows become the fields you can put on a chart. Each source arrives with its own dimensions, its own metrics and its own definition of a click or a conversion.
The same word carries a second, closely related meaning in commerce platforms. In a Meta catalogue, a data source is whatever populates the products: an uploaded or scheduled product feed, the pixel firing on your website, an app that reports events, or a partner platform such as a shop built on Shopify or WooCommerce connected directly. In both senses the principle is identical — something outside the tool supplies the rows, and everything you build sits on top of it.
Why the data source matters
Every number downstream inherits its accuracy, its definitions and its permissions. A chart cannot be more correct than the source feeding it, and a metric renamed in the report still means whatever the platform meant by it.
Permissions deserve particular attention. If a source was connected using a freelancer’s personal login, the report stops working the day that account is removed, and the client owns nothing. In a catalogue, the choice is just as consequential: products created by the pixel and products from a feed can describe the same item differently, and the platform then has to decide which version to advertise.
Common mistakes with data sources
Connecting two sources that cover the same ground and letting both feed the same report or catalogue. The result is duplication that looks like growth. In a catalogue this is the common cause of the same product appearing twice with different prices.
Assuming two platforms mean the same thing by the same word is the other. A conversion in an ad platform and a key event in analytics are counted under different rules, with different windows, and putting them side by side without saying so misleads everyone who reads the page. Leaving a source connected but never checking whether it still refreshes is a quieter failure, and reports can sit on stale numbers for weeks before anyone notices.
How to act on it
Connect every source with an account the business owns, and write down what is connected, at what scope and under whose authorisation. Name each source so a reader knows which account it came from without opening the settings.
Decide which source is authoritative for each measure — one place for spend, one for revenue, one for enquiries — and let the others be supporting detail rather than rivals. Before any report goes out, check one number in each source against the platform it came from; that single habit catches most of what goes wrong with dashboards and reporting. Where several sources have to appear on one chart, join them deliberately as a blended data source rather than stacking them and hoping.
