With the growing market for mobile applications, Google has decided to link their analytics to that of websites. For this reason, in mid-2019, it created a beta version of App + Web connecting these two platforms. This was the forerunner of Google Analytics 4, which was released in the fall of 2020.
Google announcement: “As of October 14, 2020, when creating a new service, Google Analytics 4 (formerly known as the Application and Web Type Service) is set as the default. Universal Analytics refers to the previous generation of Analytics.”
Earlier versions of Google Analytics have always been improvements to website analytics. GA4 is built on the new data model which is based on the Firebase mobile application platform. That’s why GA4 isn’t a newer version of Universal Analytics; rather it’s something completely different.
Data Model
The most important change is the data model both platforms are built on. Universal Analytics has a session-based data model where different types of hits are sent to Google Analytics. We can imagine individual hits as page views, events or transactions. On the other hand, all data sent to Google Analytics 4 is an event (event_based).
- The main difference between the two models is that there is a session in the middle of session_based model; therefore, a significant portion of the reports in UA are focused on sessions. Many dimensions and metrics are based on sessions, such as Sessions with Event, Sessions to Transaction or metrics % New Sessions, Number of Sessions per User, and Average Session Duration.
- On the other hand, the event_based model provides better information about how users behave on the web and what they do based on clicks, scrolls, and other interactions.
The UA model works with dimension decomposition according to scope into User, Session, Hit and Product scope. Dimensions and metrics from different ranges cannot be combined with each other in UA reports. GA4 does not allow you to change the range of dimensions — everything is Hit scope, or Event scope if you prefer.
Nevertheless, there is the possibility to have your own dimension for a specific user. To do this, User Properties are used to assign parameters to users such as language or location. In addition, in e‑commerce, we are used to using the Product scope dimension. In GA4 it is possible to use the so-called Item Parameter in a similar way with (item_category, item_category_2, …)
GA4 Session Definition
The arrival of GA4 is sometimes seen as the end of sessions in web analytics. However, this “end” is not yet finished. In GA4 we can still find the session metric, but it is simply not as emphasized as in UA. It ceases to be the cornerstone of reports and its definition also changes a bit.
Let’s Recall the Definition of an Event
It is common for both systems that the session starts the moment the user arrives to the website (or opens the application). The difference comes in how the session is ended.
There are two options in UA of how to end a session. The first option is expiration after 30 minutes of inactivity or at midnight. The second option is if a user comes to the site through a campaign, leaves, and returns through another campaign.
In GA4, there is only one way to end the session — after 30 minutes of user inactivity. Unlike UA, a new session is not created when the source of the campaign changes during the session. Therefore, you should be aware that you may find fewer sessions in GA4 reports over the same period than in UA.
“The global site tag automatically measures several events, for example: first_visit, session_start, user_engagement, and page_view.”
Main Advantages of GA4
More efficient measurement — In UA, each event is sent to GA as a separate request. GA4 took over the method of sending data from mobile applications, when the internet connection was not yet good. Requests do not have to be sent right away; they can wait for a few seconds, and if more than one event happens, they will be sent all at once. This helps reduce the number of requests sent to Google Analytics servers and reduces the load.
Unlimited data collection — Some large sites may run into limits of UA that can handle a maximum of 10 million hits per month. Therefore, it may be necessary to purchase a paid version of GA360 with a limit of 2 billion hits per month. As of now, there are no limits for GA4. But that does not mean that there won’t be a premium version in the future.
Cross Device (User Identification) — Better identification of users based on User-ID, Google signals and then the device. This method first determines user ID (User-ID) used to identify the logged-in user, e.g. after registration in your online shop. If a User-ID is not found, information is searched in Google Signals (if available). If no information is found either, the device ID (Client-ID) is used to specify the user, which is a cookie for sites or an ID instance for applications. Therefore, users may experience better deduplication. This is why you can find fewer users in GA4 over the same period than in UA reports.
Increased sampling rate in advanced reports — When selecting a larger time period, applying filters to select data, or increasing the complexity of queries, Google Analytics uses a sampling method (data sampling). The goal is to get meaningful information about a larger data set faster. However, the sampled data is no longer accurate and is based on estimates, which is a problem when we need to work with accurate data. Sampling limits in GA4 increase in advanced reports, and there is no sampling in standard reports.
Export to BigQuery — Another “big” piece of news is the export of data to BigQuery. This is a large warehouse in the cloud, where raw data is stored. In addition to becoming the owner of this data, BigQuery offers you the option of advanced analysis, such as your own attributions or long-term customer values. Previously, to export from UA, you were required to have a custom connector that needed to be programmed. GA4 has a free connector and it is as easy to connect as on Google Ads, for example.
Advanced forms of analysis in the GA4 interface — Analysis Hub is a tool for creating advanced reports directly in the GA4 console. You can work with higher-level data using already-prepared templates or creating your own reports and visuals. This is a tool that was previously only available in the paid version of GA360. In GA4, it is now available to users for free.
Anonymization of IP addresses — Because the IP address is considered personal information from GDPR point of view, it is the responsibility of each site owner to protect the privacy of site users by anonymizing the IP address. In GA4, this function is set by default. In UA, it was necessary to modify the configuration of the measurement code.
Automatic measurement of basic events (Enhanced measurement) — Basic measurement code of GA4 can measure not only Page_view but also a number of other events without having to implement the measurement additionally. Using a simple switch, it is possible to measure directly in the GA4 console:
- Watching videos on YouTube
- Outbound clicks
- File Downloads
- 90% Scroll Tracking
- Site Search Events
GA4 main disadvantages
Missing Views — So far, we’ve been used to working with multiple data selections in UA. For example, analytical, backup, or test data selections were common. For the time being, there is no option to adapt and filter individual data selections in GA4 and we can work with only one.
Missing advanced filtering option — Advanced data filtering before processing in GA4 is absent. It is now possible to filter only internal and developer traffic.
Missing link to other Google services — Links to other Google services are directly available only for Google Ads and BigQuery. For other services such as Google Search Console, Optimize, etc., there is no link.
Missing Referral Exclusion List — Referral Exclusion List functionality was added this spring. Until it was possible to exclude your own domain from referrals, there was a big problem due to conversion attribution. Many conversions fell under the source of the site’s domain, and this made it virtually impossible to evaluate conversions in a way that made sense.
Missing custom channel grouping — GA4 reports do not provide the ability to prepare custom channels using rules as in UA.
As other disadvantages we can mention missing known preset reports, which we know from UA. Development is still underway and some things may stop working, bugs may occur and the documentation is not yet finished.
What should you remember about GA 4?
Although Google has announced the end of Universal Analytics development, support is not ending and is likely to continue for several months and possibly years.
GA4 works on a new data design that is different from UA and is called Event-based data model. In a nutshell, this means moving away from a session and moving on to the user and events. Data from the original UA cannot be transferred to GA4. For this reason, the measurement is set in parallel. We recommend that you start measuring as soon as possible so that you have historical data in the future. So far, it is not recommended to use GA4 as the main (primary) data source, but to use mainly its new functionalities (new metrics, Machine Learning, Analysis Hub).
Google has stated that the development of GA is already directed exclusively at GA4, but it is not a finished product and is still in development. Therefore, it is necessary to take into account that bugs often occur and many of the things we know from UA are still missing in GA4.
Over time, new benefits and reasons for switching to a new measurement in GA4 can be expected to emerge. If you need help with the initial consultation of your transition to GA4 or set up duplicate measurements, give us a call.
Interesting fact: Why is the new version called GA4?
Because this is the 4th development version, which follows chronologically.
(Thanks to Honza Tichý for pointing out the inaccuracy)
The three previous versions:
- The first version is Urchin, which was acquired by Google in April 2005
- The second version is the 2007 Classic Google Analytics (ga.js)
- The third version is the Universal Analytics + SDK (analytics.js) from 2012
