In March this year, the world of web analytics undergone a real earthquake. Back then, Google announced that the most popular version of their flagship Google Analytics tool – Universal Analytics, will stop processing data at the end of June 2023. This, for some kind, forces an “upgrade” to the latest version, i.e. Google Analytics 4. Despite the fact that GA4 has been available for a long time (it was released on October 14, 2020, and was tested already in 2019 under the name App + Web), many users still prefer to use the familiar Universal version.
The “upgrade” word in the introduction wasn’t put in quotation marks by accident. Due to the number of differences, Analytics 4 can be treated as a new tool, not as a natural extension of UA. The main and most important difference is the data collection model. In Universal Analytics, it’s based on hits sent to the system – it could be, for example, starting a session, going to a subpage or a custom event configured in Tag Manager. Overall, hits can be divided into four types:
- Related to the page URL,
- Related to events,
- Related to ecommerce,
- Related to embedded social media links.
In Google Analytics 4, the data model is based on events that are triggered by an interaction with a website or application. We don’t distinguish between types of events, so it will be both going to a place on the page marked with a different URL, clicking on a properly tagged button in GTM, or purchasing a product.
You can see another big difference by looking on the tool’s interface. Universal Analytics has more extensive predefined reports in which user can view the data that are most interesting for him. However, this doesn’t mean that it’s more difficult in GA4. In this case, Google has opted for the possibility of greater customization – the exploration section allows you to create a variety of reports from the available data.
Analytics 4, as its previous name (App + Web) indicates, is also used to measure traffic in mobile applications. User interactions with both the website and the application can be measured and analyzed in one property.
Conversions in Universal Analytics can be divided into goals (visits to a specific subpage, duration of a session, number of pages visited in a session or event), with a maximum limit of 20 per view, and ecommerce events. In GA4, several events are set as conversions by default, e.g. purchase (for both website and app), first_open, in_app_purchase, app_store_subscription_convert, and app_store_subscription_renew (app only). In addition, we can define up to 30 events per service as a conversion.
The difference in favor of GA4 is codeless tracking of fundamental events. In the Universal version, without advanced configuration, the tool only tracked pageviews on the site. Analytics 4, thanks to the enhanced measurement function, can automatically collect events such as scrolls (when user arrives to the point of the page where its part lying 90% of the vertical depth is visible), interactions with videos or clicks on outbound links.
The changes also reached one of the basic metrics in Universal Analytics, i.e. Bounce Rate. It has been redefined and standardized.
In UA, we understand bounce rate as the percentage of sessions that ended without any interaction with the page (interaction is moving to the next subpage or an event). This has its downside, e.g. when a website collects many custom events, it’s highly probable that the Bounce Rate will be lower than if they weren’t collected (assuming that in GTM the parameter “non-interaction event” is set to ” False”). As a result, in Universal Analytics it’s hard to compare Bounce Rate between different properties.
In Google Analytics 4, bounce rate can be found in the exploration section. It is defined as the percentage of sessions that weren’t engaged . We understand an engaged session as meeting at least one of the following conditions:
- Lasts over 10 seconds,
- During its duration user makes a conversion,
- At least 2 subpages / screens have been displayed.
The GA4 bounce rate is the inverse of a newly created metric we call engagement rate. Engagement rate is equal to the share of engaged sessions involved in all sessions.
Reffering to engagement , the next introduced metric is average engagement time. It’s total duration of periods during which users were engaged (i.e. there was a website or application on his screen) divided by the number of active users. This is a big improvement in relation to Universal Analytics, where the time spent on a subpage is treated as the difference between entering it and moving to the next subpage (even if the user was actually browsing a different tab at that time).
One of the loudest topics in web analytics is privacy. In February 2022, Austrian Data Protection Authority raised its concerns about possible violations of the GDPR due to the use of Google Analytics, and a similar statement was made shortly after by the French CNIL. There are several changes in Analytics 4 compare to UA to prevent this from happening in the future. User IP addresses will be automatically anonymized, European users’ data will be stored in Europe, and an additional functionality is the ability to block Google Signals and collecting location or device data in specific regions.
The previously mentioned Google Signals helps us with the so-called cross-device tracking, i.e. user tracking on different devices. GA4 offers greater opportunities in this respect, thanks to a unified model for collecting data from the web and applications. The user is recognized first by the User-ID function – if an ID is not assigned, Analytics uses data from Google Signals – if it’s also not available, the device ID is used. This makes it easier to understand the customer journey from first contact with the product to its purchase.
Developing machine learning algorithms have influenced the latest version of Google Analytics. Predicted segments are new possibility. With the help of this functionality, we can examine the behavior of users who, with the likelihood indicated by us, will make a purchase or resign from activity on the website, as well as those who, according to the forecast, will be responsible for the highest revenue within the next 28 days. With this data, you can predict customer behavior and tailor your marketing campaigns to focus on high-converting audiences.
BigQuery is a tool for analyzing and storing large datasets. Exporting raw data from Google Analytics to BQ was previously reserved for owners of GA360 – the paid version of Analytics, but now it’s also possible in GA4. Limited data export is free of charge, but payments are charged after exceeding 1 terabyte of data processed in the analysis or 10 gigabytes stored in memory. However, keep in mind that there is a daily data export limit of 1 million events. Larger enterprises with high traffic to their websites and mobile applications should consider switching to the GA360 if this limit may not be sufficient.
Disadvantage of GA4 is the time-limited ability to store data. The maximum period in Exploration section is 14 months, which can be a problem when creating, for example, annual comparisons. Long-term analysis can therefore be another (but not last) reason to consider purchasing Analytics 360, where data can be stored for up to 50 months depending on user settings.
Another change is that at the time of writing this article, there is no functionality to create views in Analytics 4. In UA, it was good practice to have at least a few views per property (e.g. a main view that collects filtered data in the chosen way, but also a test view to configure new filters and a view that collects all site data). Here, however, GA360 comes in handy as well – thanks to this tool we get the possibility to create a subproperty in which we can filter the data we are interested in from the source property. This functionality works in a very similar way to the views known from Universal Analytics.
GA360 has many more extensions compare to GA4 that can be crucial in performing comprehensive analysis of complex websites and applications. Paid service allows us to increase the limits – both when it comes to data collection, reporting, storing or exporting to BigQuery. You can learn more about Google Analytics for large enterprises and corporations at fullstackexperts.eu