Data Layer

To collect information from a page, analytics and advertising tags often look at the page's visible HTML: the text of a button, a price field, a form heading. When the design changes, those links break and measurement fails silently. A data layer removes that fragile link: the site writes important events and values to a separate structure under agreed names, and tags read information only from there.

On the web, the most common implementation is the window.dataLayer array used by Google Tag Manager. Events are added to that array with names such as "form_submitted" or "product_viewed" and their parameters. Consent management also starts in this layer: the user's consent status is passed on before any tags fire. A robust data layer is a prerequisite for reliable decisions on the attribution model and on conversion rate optimisation.

• Measurement plan: A document that sets out which business question is answered by which event.
• Event and parameter names: Consistent, documented naming conventions that every team uses with the same meaning.
• Data structure: Which fields carry page, user, product and transaction information, and in what format.
• Consent status: Passing the user's cookie and data processing consent on to the tags.
• Versioning and quality control: A record of changes and testing before release.

Consider a B2B site with an enquiry form. Its analytics tool counts the form only when a "thank you" page loads; when the form is redesigned, that page is removed and conversions drop to zero overnight. The advertising platforms lose their learning too.

With a data layer, when the form is submitted successfully the site writes a single event containing the form type, the related service and the language. Analytics, advertising and CRM tags all listen for that event. However much the design changes, the event stays the same, so measurement is preserved.

The build works backwards from business questions: first decide which decisions need which data, then write the list of events and parameters into the measurement plan. The development team adds the events to the code according to that plan, and the tag manager is configured with tags that listen for them.

Testing uses the tag manager's preview mode and the browser's developer tools to check that each event fires at the right moment, with the right parameters and only when consent has been given. After launch, alerts are set up for sudden drops in event volumes.

The data layer is the structure through which the site describes what is happening; a tag manager, such as Google Tag Manager, is the system that reads that structure and decides which tool runs when. A tag manager can be used without a data layer, but information is then scraped from the page and remains fragile. The data layer is the source; the tag manager is the distributor. Getting this separation right is the foundation of advanced analytics work.