What is Google Analytics 4 (GA4)?

Google has caused a big stir in the world of web analytics with the announcement Google App + Web properties.

Google Analytics 4 (previously App + Web) is a completely new platform with a separate underlying architecture that promises to take web analytics to the next level, rather than an update to, or feature of, Google’s existing ‘Universal Analytics’.

In this article, we will highlight the key developments of GA4, how it differs from Google Analytics, and an overview of how the platform can be used.

Why you should be excited about Google Analytics 4 properties

This move by Google is a significant and bold one; given the number of websites out there that have spent a lot of effort and time implementing the current Universal Analytics, the introduction of GA4 will impact a lot of businesses.

Despite this, there are a number of features of Google Analytics 4 that will benefit businesses that we feel people should be excited for.

Truly cross-platform reporting

While Universal Analytics has the userID functionality that lets you report to an extent across platform/device, it has some significant flaws:

• Tracked hits only report under a special ‘User ID View’, and visitors need to be logged in to be recorded. Most site visitors are anonymous, especially when there is not a logged-in function. Even when there is, users tend to only log in if they are making a purchase
• Most implementations would track their app and website data into different web properties. Only GA 360 customers have the luxury of creating a roll-up property to hold all traffic.

Within GA4 properties, Google has introduced the concept of ‘streams’. You can now create multiple streams of data within the same underlying dataset/property; these streams represent your different applications, so you could create separate streams for your website, iOS app, and android app. Eventually, we expect other applications perhaps such as point of sale software or CRM to be available (although the functionality is not there to do this just yet, unless of course, those applications run Firebase).

**streams from non-Firebase SDK/ web sources will only be available if or when Google introduces measurement protocol functionality similar to that within Universal Analytics

Simpler, event-based, architecture

With GA4 properties, there are only two concepts to grasp: users and events. Rather than collecting data on sessions and hits within each session (like in Universal Analytics), GA4 collects and records events, each of which can have up to 25 custom-named parameters associated with it. This is a much simpler architecture that will be simpler to understand, plan, document and implement.

One final benefit to the new architecture is the greater ability to customise, rather than having to shoehorn all events into the Event Category/Action/Label/Value framework that Universal Analytics provides.

Faster, simpler and more flexible reporting

GA4’s simple architecture means simpler reporting and more flexibility than reporting in Universal Analytics. Interpreting and analysing the tracked data should also be easier than in GA.

Additionally, the more streamlined dataset should allow for quicker access to the data, and Google has promised that the new framework will provide lightning speed return of data to reports, and a level up from what is currently available in Universal Analytics.

One big selling point of the new Google Analytics 4 properties is that you get access to the ‘Advanced Analysis’ tool, which has historically only been available to Google 360 customers. This is a very powerful exploratory environment in which you can produce reports, visualisations, pivot tables, custom funnels, segment overlap and path analysis reports.

BigQuery export

GA4 gives users the facility to set up and export all web and app data directly into BigQuery. This hit level data opens a whole host of opportunities, which were previously only available to GA360 customers. Some of the benefits of having your data in BigQuery include:

• Join analytics data with other data sources, such as POS or CRM data
• More powerful reporting (limited only by your ability with SQL)
• Run machine learning projects to estimate things such as lifetime value or create audiences.

As the world of data and analytics evolves, we are seeing more companies create data warehouses where they can hold multiple sources of data together, enabling them to remove silos and analyse and report on data more accurately. Being able to include web and app data within these data warehouses offers great potential to the entire industry, not just those large enterprises that are paying for the paid tool.

Platform-specific features: apps

Those that have had Firebase installed in their apps for a while will know that by simply installing the SDK, Firebase will automatically start to track a set of useful events out of the box without the need to implement any further code. Some examples of events which are tracked in App + Web include screen_view, session_start, app_update and first_open. A full list of events can be found here.

To get the full benefit, you will need to plan and code specific events for your particular app within the app.

Architecture and data model

We’ve already mentioned the fact that the new App + Web properties uses a simpler event-based model, but wanted to detail this further by highlighting a key difference between the underlying data schema of GA4 and Universal Analytics.

Note – the full schemas can be viewed here and here.

As you can see, the Universal Analytics data schema looks at sessions whereas App + Web looks at events. While one of the fields in Universal Analytics in each session is a nested table of ‘hits’, the central concept is still the session; as Google processes the session, it derives and aggregates certain data and stores it against this session.

With App + Web properties, the schema is just a flat table of events. Sessions are still represented by an auto event called session_start but these are an event, just like any other. It waits to be seen if Google introduces any sessionisation logic, to sum up, and represent the aggregated session data. This flatter simpler data schema gives several benefits:

• It will be faster to query
• It is simpler to understand, and less likely to result in reporting errors from incorrect queries being run
• It is simpler to implement and design an implementation plan.

The key concepts that makeup App + Web implementation are discussed below.

Event Parameters

For each event, it is possible track up to 25 custom parameters to represent the meta data around that event. These can be textual (like hit scoped custom dimensions) or numerical (like hit scoped custom metrics).

There are some limitations, however; each project or property has a limit of 100 custom parameters (50 text and 50 numerical), and every parameter you register against an event counts even if it is the same parameter. For example, if you created a parameter called ‘state’ and registered it against 50 different events then this would use your entire quota of 50- textual parameters for the project, which means users need to carefully plan how events are set up to conserve this quota.

This is a clear limitation, and if Google create a premium version of GA4 it is likely that the limits would be significantly raised.

User Properties

User properties are essentially the equivalent of user scoped custom dimensions in Universal Analytics. For each project/property, you are allowed 25 user properties and they are intended to store longer-term data that describes that particular user: for example, ‘registeredUser’. Once you set a user property in a device, that value will be set against all future events until the value is altered or unset.

Current state and the future of App + Web properties (now Google Analytics 4)

While GA4 is an exciting development with huge potential for the world of analytics, it is still in beta and is fairly bare in terms of functionality. Until this point, Google has focused on the underlying framework of the platform and while they will be concentrating hard on adding all of the features and functionality you are used to with the current version of Google Analytics, it is far from ready to be classed as an enterprise tool in its current state.

We expect Google Analytics 4 to rapidly develop over the next couple of years, eventually taking over from Universal Analytics. As this happens, we will see new and improved features to appear as Google becomes ready to take GA4 out of beta and declare it as their primary analytics tool.

Google hasn’t declared a timeline or an explicit list of what they will be adding, but our guess is that these features will include:

• Enhanced ecommerce. At present this only consists of a transaction with revenue (not even products). This is far from the advanced functionality people have come to love with Universal Analytics Enhanced Ecommerce which tracks product activity all the way through the ecommerce funnel
• More Google tools integrations. Currently, Web + App has is an integration with Google Ads, but I would expect this to be developed to include the full Google Marketing Platform stack of tools
• Improved reporting and interface. The App + Web interface and reporting is a big upgrade from what was available in Firebase, but it is still in its early stages and I would expect this to be improved a lot as it currently lacks basic functionality to run and export standard and custom reports
• Measurement Protocol. This would allow customers to send tracking hits from other applications such as your CRM
• Data import. This would allow data to be uploaded in a similar manner to that within universal analytics. This data could be anything from refund data to user data
• App + Web currently lacks an API*. With the addition of the BigQuery export, the need for an API is reduced, but I would still expect them to provide something for reporting and for management of the account – *UPDATE: Google has now released an API for GA4 properties which you can read about here.
• Attribution. I would expect them to incorporate the stand-alone ‘Google Attribution’ tool that has been in development for some time.