WHAT IS GOOGLE ANALYTICS 4
Google is currently making huge investments in Google Analytics 4 (GA4). GA4 was previously known as the ‘App + Web’ beta and is now the default property with the intention on replacing Universal Analytics (analytics.js) in time. Universal Analytics (website) and Firebase (App) Analytics have been around for a number of years, the intersection of the information collected in these two properties was never possible since their data modelling was different.
In this post I’ll provide an overview of some of the key differences and also benefits of thinking about this upgrade early. The headline view on G4 is a move away from pages and sessions (sort of) to provide a better representation of user behaviour.
If you’ve followed Google Analytics development you’ll remember there have been three phases to the product we’ve come to know since the platform was acquired in 2005:
- 2005: Google acquired Urchin and Google Analytics was born.
- 2009: Transition from Urchin to classic Google Analytics (ga.js)
- 2013: Transition from classic Google Analytics to Universal Analytics (analytics.js)
- 2018: Universal Analytics is centralised to work across Google marketing and analytics tools (gtag.js)
How is GA4 different to universal analytics
The main difference introduced with GA4 compared to previous version of Google Analytics is the event driven data model.
The data model for universal analytics is largely session based, with a pageview sent on each screen load. As a user remains activity these hits will rollup into what is known as a session or visit depending on the tool used. A session will end when a break in activity occurs that is greater than 30 minutes. As the number of mobile and single page apps increased, the concept of views and sessions didn’t always fit. Even a web based ecommerce implementation is made up of a series of pageviews and events, in many cases the event is as important to the customer journey as the screen/page if not more. With the GA UI limiting an simple way to view and analyse page and events across the reporting suite the idea of a flat structure model across the reporting makes more sense.
Consider how useful a screen load is for an app that loads once and runs a series of processes such as video or gaming. When we think about mobile web browser behaviour, how relevant is the concept of a session and a 30 minute window of inactivity. With browser or app windows open in the background between short bursts of focus as users dip in and out the concept of session has moved away from how we interact with the web.
Some of the Google Analytics 4 differences
Bounce Rate has evolved
The concept of measuring user engagement from first load hasn’t disappeared although you wont find bounce or bounce rate within GA4/. The metric has evolved into ‘Engaged Session’, think the inverse of a bounced session plus some addition criteria to help improve the metric for single page websites and apps.
Bounce rate although a go to KPI for measuring your marketing campaign success, it can be considered somewhat once dimensional. If you think about a single page application or a news article, a user seeing one screen and leaving probably isn’t an illustration of their engagements. Step forward ‘engaged session’
Engaged Sessions are defined as the user doing at least one of the following during their session:
- Actively engaged with your website or app in the foreground for at least 10 seconds
- Fire a conversion event
- Fire 2 or more page or screen views
You’ll notice several new metrics in GA4 property that are built on top of this concept:
- Engagement Rate = (engaged sessions) / (sessions)
- Engaged Sessions per User = (engaged sessions) / (users)
- Engagement Time = sum(engagement time)
The new metric you’ll want to use instead of Bounce Rate is Engagement Rate.
Sessions & Channel attribution is changing
Sessions remain a confusion point in GA4, the headline being a move away from this model however Google are calculating and using sessions within the model.
Changes here are likely to present as a decrease in session volumes when compared with universal analytics, this is as result of some changes to the midnight cutover being removed. Perhaps more fundamental to your reporting, a session will no longer break midway through if the campaign source is overwritten. Based on the comparisons I’ve personally run (as of August 2021) I’m not seeing -0.5% sessions over a 30 day period.
Additionally we can expect some changes in User volumes as a result of GA4 looking for active users compared to universals total users. Based on the comparisons I’ve personally run (as of August 2021) I’m not seeing -1.5% sessions over a 30 day period.
More clarity is required on this and I’ll updated as this comes through.
Filters are not included
Filters at the view level in universal have multiple use cases across the board, the most basic being to filter out internal traffic. GA4 as it stands does not include this functionality, I for one think this is going to be a major issue for anyone looking to ensure the data they report is clean and customer-centric. (updated Aug 2021)
Future Proofing Privacy
In view of the increasing cookie restrictions and the resulting scarcity of data. Google are positioning GA4 and its use of machine learning to maintain and fill in any data gaps left by the absence of cookies.
For any machine learning program to be sufficient you’ll need a good foundation of healthy historic data, one good reason to add GA4 as a secondary tracker sooner rather than later.
Machine Learning is intended to provide forecasts and predict results such as churn (user turnover rate) and potential revenue for a given segment.
I’m not so sure on this privacy standpoint, with my analyst hat on I’d rather have the data vs a modelled view on the world. If my spend on Google is being focused by it… But for now, I’m keeping an open mind.
Previously this feature was exclusive to GA360 subscribers but is now available to everyone. You will be able to export your analytics data to BigQuery, either for storage or future use by integrating with other data sources. A use case for BigQuery is access to raw data, interesting to see how this play out with the increase machine learning and reduced actually data.
Easy integration with Google Suite
Integration with Google marketing tools is a clear focus for GA4, this could be considered a side step into a marketing analytics tool and perhaps moving away from a web/digital analytics tool. Its early days but I hope this direction of travel doesn’t impact the eventual core web analytics reporting and data management tools required for optimisation.
There is going to be a barrier to setup with this model, the implementation methodology is time consuming and not comparable with any of Googles previous protocol upgrades for Google Analytics.
I might change my mind on this in time, but at present I would advise a watching brief on GA4 and keep running Universal analytics (gtag.js) as your primary analytics solution. I have found use cases with a number of clients to push for GA4 but as yet not to be the sole analytics or primary analytics solution. If there is a killer feature in GA4 that your business would benefit from then setup for that and run in tandem with universal, running regular comparisons to understand the differences in your data sets.
If you are an app focused business then this view changes and I’m assuming you already use firebase and have made the upgrade to app + web (GA4) to measure behaviour.
The change in browsing behaviour and growth in mobile web can’t be denied, but a paradigm switch looks like a move to the middle ground for web analytics and with that compromise against what analysts and marketers have become accustomed to using since 2009.
If you’re thinking about starting on the GA4 journey or need advise around universal and the switch up to gtag.js do get in touch.