The GA4 Landscape – 5 things to look out for
Over the course of the last two years, Accord has successfully migrated over 150 accounts to GA4 and through this process we have discovered some key factors for marketers to be aware of when setting up and using this analytics service.
Here we’ve listed five fundamentals; get these right and you’ll have a more robust and accurate set of data to report and analyse.
Data retention
By default, GA4 sets the retention period of user-level data to two months. Whilst this doesn’t affect standard aggregated reports, it does affect explorations - and two months data is not much use when trying to analyse year-on-year or wider seasonal trends. It’s, therefore, recommended to set the period to the maximum of 14 months.
Filters
Excluding internal traffic can help clean up your data and make reporting more accurate. Whilst simple exclusion filters were used in Universal Analytics (UA), in GA4 there are two steps required. Firstly, internal traffic needs to be defined within the data stream settings and then, secondly, the internal traffic filter needs to be activated within the data settings > data filters.
Custom dimensions
GA4 is capable of tracking many events, such as outbound clicks, and video plays out-of-the-box through enhanced event measurement, which is brilliant. However, as standard, only the basic event names will be tracked. For example, if you want to see which video was interacted with, a custom dimension for video_title needs creating. Please see below for a list of custom dimensions to consider setting up to make the most of the data available:
4. Reporting identities
GA4 has three different methods to associate events with users:
Device-based only utilises cookies to identify users.
Observed uses cookies, Google signals and User ID to help tie up cross-device journeys.
Blended uses cookies, Google signals, User ID and modelled data to help estimate untracked user activity.
The system will use the best method available for the account as default, but you can freely switch between reporting identities without making any permanent impact on data. A word of caution, however. Data is only exportable through the API/BigQuery using the device-based identity, so if the interface is using the observed or blended identity, expect a difference in data between the interface and the export.
5. UA export
Last but not least, don’t forget your historic UA data. UA holds valuable information that goes back many years, including pre-pandemic. This will be available to access until July 2024, so set a target to export the key dimensions and metrics by the end of the year. This could include audience, acquisition, behaviour, or conversion level data by month/year. This can be exported by creating custom reports, or via an API connection. Top tip – watch out for sampling when exporting longer periods or larger datasets; export in smaller chunks or to higher granularity if this is occurring.
As part of our data portfolio of services, Accord provides deep-dive GA4 audits. Get in touch with Dan Ward, Head of Performance, to find out more.
Dan Ward
Dan is Accord’s Head of Performance and GA Guru. Get in touch if you want to pick his brains on all things marketing performance.
