Chapter 2
In some ways, it feels like the Google Analytics we all have been so comfortable with (Universal Analytics) stayed remarkably consistent since its launch in 2013. Core concepts—including conversion rate, goals, session duration, and bounce rate—formed the base of many marketing and analytics conversations for the past decade. However, behind the scenes, Google Analytics rolled out a suite of changes over the past 5 years that culminated in the launch of Google Analytics 4.
In this chapter, we’ll cover Firebase Analytics, the Global Site Tag, and the shifting privacy landscape, which all shaped this new version of the tool.
free migration assessment
Search Discovery has designed an assessment to help you prepare for your migration to Google Analytics 4.
Get Your 5 Minute Assessment TodayRecent Evolution of Google Analytics
Google Analytics 4 is the first major evolution of the product since the 2013 release of Universal Analytics (analytics.js). GA4 (based on the Global Site Tag, gtag.js code) is an entirely new event-based data model built off of the analytics of the Firebase app development platform.
Even though users logging into Google Analytics did not see any major changes in the product or reports for many years, quite a bit of product development has been underway since 2017. Many of us watching the early developments wondered, “where is Google headed with all of this?”
Timeline:
2017
Google released ‘Google Analytics for Firebase’ and Google announced the timeline for sunset of its original Google Analytics SDK for app analytics tracking.
Additionally, Google announced the rollout of the Google Site Tag (gtag.js) as the standard tag for sites not tagged with a tag management solution like Google Tag Manager.
Additionally, Google announced the rollout of the Google Site Tag (gtag.js) as the standard tag for sites not tagged with a tag management solution like Google Tag Manager.
2018
Google Signals launched. By activating Google Signals, Google provided a handful of cross-device analytics reports in GA to better understand cross device traffic and acquisition for users. The reports were generated from the subset of users with a Google account that had not opted out of ‘ads personalization’ (the ability for Google to target ads to users across multiple devices). Activating Google Signals also activated the ability to target ‘audiences’ across multiple devices with audiences created in Google Analytics and imported into Google Ads.
2019
Google released a beta version of GA4 as ‘App + Web Properties’ in July. Three months later, Google deprecated the Google Analytics SDK for app analytics tracking (which used Universal Analytics) so that all mobile app tracking would migrate to Firebase Analytics (which enabled the new ‘App + Web’ properties).
Oct. 2020
Google renamed ‘App + Web’ as Google Analytics 4 and announced it would become the default platform for Google Analytics moving forward.
Oct. 2021
Google officially announced the rollout of the new Analytics 360 (the enterprise version), with GA4 as the platform for the future of the paid Google Analytics product.
Mar. 2022
Google announced that Universal Analytics properties will stop processing data on July 1, 2023 (360 properties have an additional 3 months until October 1, 2023), with access to reports in Universal Analytics for ‘at least 6 months’ after data stops processing.
May 2022
At Google Marketing Live, Google announced upgrades to gtag.js as ‘The Google Tag’ (or ‘One Google Tag’), which allows adding/managing advertising and analytics products from within the admin settings without changing website code.
So, here we are with a new version of Google Analytics, GA4. The timeline provides details about ‘what’ Google has been releasing over the past 5 years, but it’s also helpful to understand ‘why’.
New Google Analytics—Built for the Current Privacy Landscape
Privacy Regulation
As a result of increased privacy regulation (GDPR in Europe since 2018, CCPA in California since 2020, among others), more sites are required to give users the option to opt out of a variety of marketing and/or analytics tracking.
We will discuss privacy regulations in depth in chapter 10, but, in this chapter, it is important to point out that when GA4 was in development, a large portion of the industry was moving toward ‘opt-in’ analytics policies, which would spell a future of a much less complete view of site usage.
Browser Changes & Cookies
In response to individual concerns and the shifting privacy landscape, most browsers and platforms (in particular, Apple’s iOS) have integrated increased privacy settings that often eliminate third party cookie tracking and limit the lifespan for first party cookies.
Google Analytics sets a first-party (same domain as the site) cookie to recognize return users, but a shorter lifespan impacts ‘user’- level metrics. This causes total users to be inflated. When returning users’ cookie is no longer available, the count of new/return users is incorrect because more users appear ‘new’ with the shorter cookie lifespan, and marketers have a reduced ability to attribute campaign history to users. (If they don’t return within the cookie lifespan, prior campaign history will no longer be available).
Previously, with a longer lifespan for the GA cookie, Google Analytics had a better view to multitouch attribution by leveraging a user’s campaign history. Shorter cookie lifespan hinders this for impacted browsers, and it decreases the ability to attribute a returning user to a prior campaign.
Google’s Use of Modeling to Address These Challenges
To compensate for data lost to browser/cookie limitations and users opting out of tracking, Google will include modeled conversions in GA4 to augment data collection with estimated conversions impacted by browser limitations (read more about conversion modeling here). It will also include modeled events for users opting out of data analytics cookies with ‘consent mode’ through Google Tag Manager (this feature is called “behavioral modeling”). This ‘blended’ (use modeled data) setting is now the default setting within GA4.
Additional Privacy Features of Google Analytics 4
In chapter 10 we will discuss the legalization dispute in Europe regarding Google Analytics in detail along with an exhaustive list of privacy features that are now available in GA4, but here are a few highlights that differentiate GA4 from prior versions of Google Analytics. First, IP addresses are not saved with GA4 data collection (previously, in Universal Analytics, IP ‘anonymization’ was an option but not the default), and GA4 goes a step further to not ‘log or store’ a user’s IP address (a geo-lookup is performed, but the data is not stored). GA4 is also integrating region-specific data storage to comply with EU data privacy regulation, i.e., GA4 will collect and store EU data in the European Union.
Given this new reality, user identification in GA4 is built to rely on cookies as a last resort for identifying a user.
Google Analytics 4 will first look for a logged-in user identifier (if captured), then a Google identifier (if Google Signals is enabled for the property), and finally the device-level cookie identifier if neither of those is available. Google Signals is the user’s Google identifier, which is known to Google for users logged into Chrome, on an Android device, or logged into Google products on another browser/platform and available for cross-device and cross-browser identification, as long as the user has not opted out of ‘ads personalization’ in their Google account. Previously, enabling Google Signals for analytics granted access to a subset of cross-device reports, but now it is fully integrated into the foundation of data collection and reports for GA4.
GA4 vs. Universal Analytics - Key Differences
Google Analytics 4 also introduces several new concepts in addition to the privacy features mentioned above. Here are the key differences that users who are familiar with Universal Analytics will notice when they begin using GA4.
Event-Based Data Model
You have heard the phrase ‘event-based data model,’ but what does that mean, exactly? The data model is built on all data collected as ‘events’ and associated with a user. Even pageviews and screenviews (for apps) are a type of event.
Sessions are still available in the reports, but they are much less prominent, with much more of an emphasis on events and users. App tracking has featured this measurement model, as ‘sessionization’ for app activity is much less straightforward than it may be for a website. Additionally, with more sites built as single page apps, less emphasis on ‘pageviews’ makes sense when the task of designating which content should be considered a ‘page’ is essentially up to the analyst designing tracking for the app.
And, clearly, a major benefit of the event-based data model is the ability to capture data for both a web and an app in the same format, and to be able to send the data to the same GA4 property for comparative analysis across the two platforms.
Get a free GA4 Migration assessment to help understand your business’ unique benefits and challenges.
Data collection, Universal Analytics vs. GA4:
Cross Device Identification
The new ‘blended’ identity provides a more accurate estimate of true ‘users,’ as it is closer to identifying ‘people’ than the prior reliance on the cookie as the main identifier. GA4 can identify users across multiple devices or browsers who either log into the tracked site or are identifiable by Google from their Google account. Attribution reports can now integrate cross-device user identification for much more robust campaign attribution.
GTAG.js Features
One key feature of the current version of the tracking code (gtag.js) is the ability to dynamically customize the code for each property based on the data stream settings. For example, GA4 allows us to adjust cross domain tracking from within the admin settings, where previously it had to be set up in the code (or in the tag management solution).
Additionally, for sites with a simple implementation, additional events such as offsite clicks, document downloads, YouTube video engagement, and scroll tracking can be turned on from within the settings for a data stream, literally with the click of a button.
However, if additional attributes need to be collected with the ‘enhanced measurement’ events, these should most likely be implemented manually to allow for custom data collection rather than through the data stream settings.
More Sophisticated Engagement Metrics
GA4 has integrated much improved engagement metrics compared to the built-in engagement metrics in Universal Analytics. The new ‘engaged session’ uses a blended criteria of a session “that lasted longer than 10 seconds, or had a conversion event, or had two or more screen or page views.”
As GA4 rolled out, this engagement metric completely replaced prior engagement metrics (bounce rate, session duration, pages/session). Google has slowly started to reintegrate several of these classic engagement metrics to GA4, however there are many reasons to start to migrate to the newer engagement metrics released with GA4.
Several of the new metrics are based on the combined engagement metric—engaged sessions, engagement rate, and engaged sessions per user. Additionally, average engagement time is a big improvement over both time on page and session duration from Universal Analytics, both of which were based on comparing various timestamps to book-ended events and had challenges providing insight or reliable data for content engagement.
Other Notable Differences of GA4
A few differences have jumped out to clients as they become familiar with the new data format:
- Default data retention for standard properties is just 2 months (for explore custom analysis, with an option to choose 14 months. New Analytics 360 offers retention options up to 60 months.)
- The concept of a session ‘goal conversion’ (so fundamental to a Universal Analytics implementation) no longer exists. Conversion events are goals marked as ‘conversions,’ but they track each time they occur, so the baseline numbers can differ significantly from Universal Analytics (where goal conversions are limited to once per session).
- Default reports launched without any out-of-the-box conversion rate calculations, just conversion totals, but, given user feedback about this key KPI, Google has announced conversion rates will be returning.
- The concept of filtered Views no longer exists. All data flows into a single GA4 property, and filters for excluding data are limited. Analytics 360 clients do have the option to create ‘subproperties’ and ‘roll-up’ properties.
- Event batching is another difference. Similar to app events, events are ‘batched’ client-side and, except for conversion events and browsers operating in debug mode, sent in groups, rather than immediately with each hit.
- GA4 includes much fewer standard reports, balanced with much greater flexibility for custom analysis in the ‘Explore’ section.
Changes to the Enterprise Version of Google Analytics 4 (New GA 360)
Certain 360 Features Rolled Out for All Properties
Several features previously only available for enterprise GA360 customers are now available to all customers in GA4:
- Data Driven Attribution (a data model to assign weighted contribution across multiple campaign touchpoints) is now set as the standard attribution model within GA4.
- BigQuery export is now available for all properties. Data can be accessed in Google Cloud console directly for analysis, or it can be passed through to another database.
- Additional GMP integrations (DV360 and Ad Manager) are now available for all properties.
These features provide strong incentive to migrate to GA4 prior to the 2023 deadline.
Lower entry pricing model for 360 subscription
Many medium-sized organizations will find the pricing model for 360 features more enticing, with entry-level pricing quite a bit lower than what it was for Universal Analytics.
The 360 level provides more custom data collection, a much higher number of Audiences that can be created, longer data retention and higher API and BigQuery export quotas.
GA4 is the future of Google Analytics. Already, there is much to embrace—the new identity methods, data modeling, privacy upgrades, and a new pricing model for the 360 product. The product is still evolving, and we’re excited to see what additional features will be built into the tool over the next few years
