Google recently announced the rollout of Google Analytics 4, the future version of its analytics platform. Our analytics team has compiled a few key initial questions (and answers!) about GA4, with more to come.
- Do I need to start over on the analytics implementation we just did recently?
- I heard this version of GA4 previously referenced as App+Web. Since our organization doesn’t have a mobile app, do I still need to migrate?
- Does the new GA4 have any sampling?
- What happened to engagement metrics such as bounce rate and time on page/site?
- How is the new engagement time metric calculated?
- Can I use GA4 to build audiences for remarketing across the Google Marketing Platform, Salesforce Marketing Cloud, etc.?
- Is there a paid (360) version of GA4 or is there only a free version?
- I don’t have GA360. Can I export GA4 data to BigQuery for free?
Do I need to start over on the analytics implementation we just did recently?
No. For organizations that have implemented robust analytics tracking using a tag management solution such as Google Tag Manager, it’s a matter of setting up GA4 tags with existing Triggers and Variables that have already been set up for UA. Hard-coded gtag implementation also requires only a simple, one-line addition of the Measurement ID for a GA4 data stream for standard GA4 tracking.
Resist the urge to simply map your event category, action and label over to GA4 and instead thoughtfully take the opportunity to rethink your new event schema. As GA4 will be the new standard going forward, it makes sense to take the time to carefully plan out your reimplementation to ensure scalable and digestible reporting.
Keep in mind that GA4 now allows for event renaming and creation in the interface, which leaves room for some trial and error if an event name or attribute needs to be changed post-implementation.
I heard this version of GA4 previously referenced as App+Web. Since our organization doesn’t have a mobile app, do I still need to migrate?
Though Google first branded this new property type as App+Web in 2019, GA4 is an overhaul of Universal Analytics and is designed for any organization to use, even those without a native iOS or Android mobile application.
We strongly recommend that organizations begin a roadmap to implement GA4 in 2021 while keeping their Universal Analytics implementation alive. For organizations that do not have a mobile app, you can still utilize GA4’s new streams feature and create a stream for your desktop site.
If an organization has an app and a website, GA4 offers the added benefit of unifying users across their app and website experiences. Organizations can choose to enable “Google signals data collection” in Data Collection settings and “By User ID and Device” in the Default Reporting Identity settings. By turning on this functionality, GA4 will first use a persistent user ID (usually captured upon authentication) to stitch together users across platforms. When a unique user ID does not exist, Google will use Google Signals — leveraging data collected from authenticated Google users who have turned on Ads Personalization — to stitch together a holistic view on multiple devices and browsers.
Does the new GA4 have any sampling?
There is no sampling for any GA4 standard reports. While Universal Analytics has limits of 500,000 sessions for a given data set within the specified date range in the property before sampling kicks in, GA4 reports do not have sampling even with additional dimensions applied. This allows users to apply dimensions and segments in the standard interface without worrying about the dreaded yellow sampling icon. GA4 also does not have an overall hit limit, as the measurement paradigm shifts over to event-based tracking, compared with the previous 10 million hits per month limit in UA.
The full data set is also available for export to BigQuery with no sampling. There is some sampling on Analysis Hub, “a collection of advanced analysis techniques that go beyond the standard reports” that 360 users have sneak-previewed. Analysis Hub is a key feature that GA4 has made available to all users for deeper ad-hoc analysis with an easy-to-use drag-and-drop interface.
Sampling does apply when the queried data set exceeds the quota for free accounts, which is 10 million events. Sampling aside, for properties with a robust implementation Analysis Hub allows for the type of flexible analysis long desired by analysts. Users can now leverage different templates such as Exploration, Funnels, Path and Segment Overlap to explore their data.
The funnel report, in particular, allows users the flexibility to define a sequence of events (page views or a user interaction) with the ability to drag and drop segments and apply filters to discover top conversion paths and friction points.
What happened to engagement metrics such as bounce rate and time on page/site?
GA4 dropped the problematic and easy-to-manipulate “bounce rate” in GA4 and instead integrated several new engagement metrics — engaged sessions, engagement rate, engaged sessions per user and average engagement time. These metrics can be found under “acquisition” rather than “engagement” in GA4. Google defines an engaged session as a session “that lasted longer than 10 seconds, or had a conversion event, or had two or more screen or page views.” Many analysts will welcome the blend of time and action into a single, simplified metric.
Google provides a bit more detail about how it measures engagement with its definition for the new “average engagement time.” Engaged time is defined as “length of time that the app was in the foreground, or the web site had focus in the browser.” This is a much more nuanced measurement than “session duration” in Universal Analytics, which simply calculates the amount of time between the first and last interactive hits without considering the amount of time a user may have switched to another tab in the browser.
To compute engagement time Google collects a parameter “engagement_time_msec” (milliseconds of engagement) along with enhanced measurement events such as scrolls, custom events and additionally the event “user_engagement” if needed. Given that user_engagement seems to be little other than a vehicle for sending the engagement time parameter, it’s surprising that Google provides details for user_engagement in the events reports. In Google GA4 documentation about automatically collected events, Google explains an event is collected when the app or web page is opened and “periodically, while the app is in the foreground.” Engagement time is then summed per session and averaged to calculate the “average engagement rate.”
How is the new engagement time metric calculated?
GA4 now collects an “engagement time” parameter (an event level attribute to collect details for reporting) along with many of the site interactive events and, when needed to collect engagement details, a new event called user_engagement. Conversely, if a user switches to another browser window (or minimizes an app) the time the app or browser window is minimized is not included as engagement time. The engaged time is used to determine whether a session qualifies as an “engaged session.”
Summing up the engagement time creates a much more accurate measurement of actual user engagement than existing “session duration” or “average time on page” in Universal Analytics. While it has been possible to collect engagement time with custom code in Universal Analytics, this new out-of-the-box metric is a welcome addition to the GA4 platform.
Can I use GA4 to build audiences for remarketing across the Google Marketing Platform, Salesforce Marketing Cloud, etc.?
In the free version of GA4, publishing remarketing audiences is expanded to publishing audiences not only to Google Ads only, but also to DV360, SA360 and CM360. With the free version you are limited to 100 audiences for each of these platforms. The upgrade to GA4 360 increases the audience limit to 400, plus advertisers will be able to publish remarketing audiences to Salesforce Marketing Cloud. Another major change is that as soon as you create new audiences in GA4 they’re automatically published to your linked Google Ads account; no extra step is required, as with the current Google Analytics.
Moreover, users can be actively removed from a published audience. Previously, once a user was added to an audience they would remain in that audience until they hit the membership duration limit or you unpublish or close the audience. For example, if you add a non-customer to a remarketing audience with a membership duration of 30 days but that user makes a purchase seven days later, in Universal Analytics they’d remain in the non-customer audience for the entire 30 days. In GA4 they’d be removed as soon as the purchase is completed. You could of course work around this by using a customer audience for exclusion or suppression in the ad-buying platform, but this new “state-based” audience really makes managing this more efficient.
Is there a paid (360) version of GA4 or is there only a free version?
At this time there is only one version of GA4 and it’s free. There is an Enterprise version of GA4 in beta that we expect will offer service-level agreements and a larger number of custom dimensions and metrics, audiences, and higher volumes for BigQuery exports, but there’s no official timeline for that to come out of closed beta. Existing GA360 customers can launch a new GA4 property which won’t count against their existing hit quota for tiered GA360 pricing.
Even though GA4 is free, it comes with some great features formerly only available with a GA360 subscription. The most valuable of these is undoubtedly the free BigQuery export. While users of standard GA already can take advantage of the reporting or management APIs, these simply don’t stack up to the depth of data available in the BigQuery export, and they require development effort. So this is a huge value!
The other major feature in GA4, formerly only available via a GA360 subscription, is custom funnel analysis and reports. In fact, not only are these now free, but they’re much more intuitive and easy both to create and view in GA4.
I don’t have GA360. Can I export GA4 data to BigQuery for free?
Organizations that have implemented GA4 can export their raw, unsampled data to BigQuery for free. Users can select to have the data exported once a day or intraday for faster access to same-day data. Teams that have selected to export their data are subject to the same active storage and processing limits as the free BigQuery tier, with 10 GB of active storage and 1 TB of processed query data per month.
Once a premium feature, now all customers can access hit-level, raw GA4 data inside BigQuery for powerful querying and inexpensive data storage on an almost real-time basis. The integration between GA4 and BQ is a simple click in the admin section once your BQ project has been created. Analysts can choose the data location and streaming frequency and data will start flowing into BigQuery almost immediately.
how we can help you.