As of July 1st, Universal Analytics has officially retired and we have finally entered the GA4 era. Since its announcement, GA4 has been the talk of the town. We spent last year learning, adapting, migrating, struggling & using GA4 for its maximum capabilities. But this transition was challenging. Since the beginning of Digital Marketing, Google Analytics has been the backbone of any tracking phenomena & has been the top choice of marketers to compile & gather data.
Since we became more sensitive around data privacy & user consent in tracking their data, we were losing so much data on Google Analytics. GA4 came right on time to help us gather this data. Apart from data model changes & a significant appearance change, the addition & refinement of machine learning capabilities is the most potent upgrade that this tool has brought.
Using some of the features in GA4, we can compensate for the lost data due to cookies & user consent. Unlike other tools, GA4 can capture both observed & unobserved data. Before we move ahead, let’s take a look at the major difference between these data streams.
Science of Unobserved Data
The major distance between unobserved and observed data lies in the distinction between gathered data and modeled data. Utilizing unobserved data is a valuable strategy for staying updated with the dynamic landscape of digital marketing analysis, regardless of the analytics tool you employ!
Back in the day, tracking users with cookies was a breeze because almost every browser happily accepted them. Here’s how it worked in the world of analytics: when someone landed on a website, a magical cookie would be used to stamp them. This cookie granted platforms like GA the power to recognize users based on their device details, location, demographics, and, a super-sticky random ID.
Now, Once that user makes a comeback to the website, GA spots the ID & immediately identifies the user. It then cleverly weaves together the user’s previous data with their new activity. Mobile apps like to join in on the fun too, but they do it in their own quirky way. Instead of cookies, they sport a unique advertising ID as their special identifier (Android and iOS both have their own versions of it).
By default, Google Analytics never tracked any information that could personally identify you. This was because gathering such data went against their terms of service. However, the definition of personally identifiable information (PII) has evolved over time, depending on how various laws and security teams have written and interpreted policies.
Nowadays, users have the power to block and choose not to let analytics tools collect their data Such as your name or contact details. In fact, for GDPR and other countries’ laws, the automatic choice is to opt out of data collection.
Make Up for the Lost Data Using These GA4 Features
Google Analytics 4 has some ready-to-use features that can compensate for any missing data. These features are super easy to use once you have tracking set up, so you can try them out and benefit from them right away.
- Data-driven Attribution
In GA4, instead of being in the Reports section, you’ll actually find data-driven attribution in the Advertising screen. The Advertising reports are pretty insightful because they offer a different perspective on your data.
In Google Analytics or Universal Analytics, the closest thing to DDA was the Multi-channel Funnel reports. These reports are quite handy as they dive deeper into analyzing conversions across multiple touchpoints and give us a complete picture of the user journey. In Universal Analytics, data-driven attribution was only available to paid 360 accounts, but now it’s accessible to everyone.
The DDA (Data-driven Attribution) model uses fancy statistics to help us understand how important a channel is in helping a conversion happen. Let’s say, for example, that the main GA4 acquisition report shows 1000 purchases credited to the Organic Search channel. But the previous interactions with the Paid Search channel might have played a huge role in convincing the user to make the purchase.
To figure this out, the statistical model takes into account all the data about how users behaved and the different paths they took before making a conversion. It then decides how much credit each touchpoint should receive. In our previous example, instead of giving 100% credit to the Organic Search channel, the credit gets divided among all the channels the users interacted with before finally making a transaction.
The visualization of DDA is located in the Advertising>Conversion Paths report:
- Predictive Metrics
We have data about what users have already done and interacted with, but what will they do in the future? It’s like trying to predict what they’ll do next. However, it’s important to note that this feature currently only applies to ecommerce and churning data.
To use predictive metrics and predictive audiences, you’ll need to set up ecommerce tracking first. Once that’s done, the Explore reports and the Audience tool are the places to be for leveraging predictive modeling.
In the Explore reports, the best way to use predictive metrics is through the User Lifetime technique. In this type of report, you can select specific metrics that are based on things like the likelihood of a purchase, the chances of a user churning, and predicted revenue. There’s a dedicated section for these metrics on the selection screen.
By analyzing the data from users who have made a purchase and comparing it to those who haven’t, the model learns patterns that help determine probabilities and percentages. When it comes to churn, the model looks at active users and those who become inactive to figure out who won’t return to your website or app in the following week.
These insights aren’t limited to Google Analytics only. You can also use them outside of the platform. For example, you can create audiences and segments to identify people who are likely or unlikely to make a purchase, and then use this information in Google Ads for remarketing purposes. If you want to quickly create a predictive audience, just go to Admin, then Audiences, click on New Audience and select Predictive.
- Behavior Modeling
Among these three features, behavior modeling has the most significant impact when it comes to machine learning because it directly affects how user tracking happens right from the beginning, starting with the identifier. This feature requires integrating GA4 with your cookie consent management tool, allowing Google Analytics to collect data from users who don’t consent to be tracked.
The collected data is actually anonymized and not connected to any specific cookie or user identifier. Instead, it’s used to analyze user-level activity based on the events recorded. This is powerful because it relies on the data from your own website or app. The behavior of users who have consented to tracking is used to train a machine-learning model that can estimate the behavior of users who opt out of tracking.
To enable behavior modeling in your GA4 account, head to Admin, then Reporting Identity, and select Blended.
Conclusion
With the retirement of Google Analytics and the introduction of GA4, we have entered a new era of tracking and analyzing data. GA4 provides features like data-driven attribution, predictive metrics, and behavior modeling to compensate for lost data due to cookies and user consent. These features offer valuable insights and leverage machine learning capabilities to understand user behavior and make predictions for future actions. By using GA4 and its various tools, marketers can gain a deeper understanding of their audience and optimize their marketing strategies.
