Give us 9 minutes to convince you that GA4 is better than UA

Yep. You read that right. Not only do we know that Google Analytics 4 is better than the outgoing Universal Analytics, but we’ll also convince you of this perspective.

As you likely know, UA is queued up to be sunset in the summer of 2023. That means digital marketers like yourself should be actively transitioning to the new GA4 platform.

Like anything new, that means you’ll need to confront change, shift some working habits, and learn some new processes. The good news is that, with GA4, you’ll be able to take your analysis to new heights (and depths). 

To get the low down on the most significant differences between the two generations of analytics tools, we sat down with Romina Henritzi, digital strategist at Bluebird Media — a Funnel partner specializing in web analytics best practices among other digital marketing services.

While you can check out her video mini-series on our YouTube channel (or watch the embedded video below), we also wanted to break down some of the highlights here.

The difference is in the data model

Universal Analytics was built on a stiff hierarchy data model of User > Session > Hit. You can think of this almost like an old nesting doll.

In the UA system, a user is basically any browser that has a UA cookie dropped “on” it. That can mean a browser on a laptop, a desktop, mobile device, etc. — regardless of whether that is the same human being logging web activity across those devices (more on this later, though).

Each user can have one or more sessions on a website. Then, each session may contain one or more recorded hits. In the UA case, a hit is any action taken on the website. This includes page views, clicks, conversions, etc. 

The user is the outermost layer of the nesting doll, inside of which is the session(s). Within each session doll is one or more hit dolls. Alternatively, you can think of it like an academic outline from your old school essays:

1. User on laptop

A. Session number 1

i. Page view
ii. Click

B. Session number 2

i. Page view

C. Session number 3

Looking at the system from this perspective, it can seem pretty straightforward and logical. However, in today’s modern web analytics space, this hierarchical model leaves itself exposed analytical holes that prevent deeper analysis. This tiered model also prevents us from applying machine learning techniques, since the data is so siloed. 

There must be a better way. 

Enter GA4

And indeed, there is a better way: Google Analytics 4. In the new data model, the siloed hierarchy is scrapped in favor of a flat data structure. Google achieves this by making the entire system event-based. 

What’s an event in GA4?

Simple. An event is anything and everything that happens on your website. It’s almost like “hits” in the UA data model, but GA4 takes the concept a few steps further. Beyond just conversions or views, an event could be the beginning of a new session or a user’s first visit to your site. 

Again, it can be anything and everything. 

So, what happens to users and sessions?

While GA4 is seemingly all about the events, users and sessions don’t disappear entirely. Rather than being forced into the nesting doll hierarchy of UA, Google Analytics 4 sets them free (in a sense). 

Now, each event carries a parameter with the session and user ID. In other words, GA4 records that “X” event was triggered during “Y” sessions by user “Jane Doe.” In this way, sessions and users help to identify each event while placing the data from each event at the forefront for analysis. 

As we’ll get into further below, this helps shift analysis toward data points that are delivering real value for your users.

For now, let’s focus on one of the most welcome benefits that GA4 delivers…

Cross-device tracking is here — finally!

Ask web analysts and digital marketers what one of their biggest challenges was while reviewing website performance, and they will likely bring up their inability to track the same users across multiple devices. 

We all know from our own experiences that we often look at the same website across multiple devices, but we haven’t been able to reflect this behavior in our web analytics. This has resulted in the double counting of users, leading to inflated traffic numbers.

The limitations of the UA database

The reason we faced this limitation is the underlying structure of UA. In the old model, the user IDs were based on the specific browser. That means the laptop version of Safari and the iOS version of Safari were each entered as new users within the UA database.

Limitations were meant to be broken

Although this user-ID-per-browser setting is still the default in GA4, you now have options. 

If you have a customer login feature on your website, you can assign this unique login ID to stand in for the user ID, rather than the browser ID. Therefore, once a user logs into your website, GA4 can track that unique ID whether it is accessing your website on a desktop, mobile device, or both. 

Eureka!

The catch

The trick is that you’ll need this customer login function to enable this feature of GA4. But before you rush off to your website development team to demand a new customer login feature, you’ll want to give it some serious thought. For instance, what value would this feature deliver? Is there a real purpose, other than web analysis, that the user will gain? How does this affect the overall user experience?

For some websites, like e-commerce, this is an obvious feature in which to invest. For more informational and educational websites, though, the justification for a unique customer login is less obvious.

You’ll also need to think about some incentives for your users to create an account (particularly in an age of increased privacy concerns). Thinking of e-commerce websites again, you’ve probably noticed many online retailers offering 10 percent off your next purchase when you register your email or create an account.

Enough about devices, though. Let’s move on to something a bit more refined.

Sessions, refined.

While Google refers to its new session logging system as a refinement, we think this is actually another game changer as we try to achieve more accurate traffic (and specifically session) analytics.

Under Universal Analytics, a new session was recorded if a user continued browsing a website through midnight. It also registered new sessions if the traffic source changed. Now, in GA4, we avoid this double counting and maintain the same session ID.

Each of these concepts can be a bit confusing, though, so let’s break down what each means and how they’ve changed in GA4. We’ll explore each through the lens of an e-commerce perspective.

Changing traffic sources

Let’s imagine that you’re shopping for a new pair of gym shoes. Like most online shoppers, you start out with some Google searches to get a sense of the different models, styles, features, prices, etc. You might check some review sites or some online stores.

During your Google search, you might open links in a couple of tabs (or, in this author’s case, like 30+ background tabs in the same window). Then, you realize that you are actually looking for some high-quality, long-distance running shoes. Something lightweight but supportive, comfortable, reliable, and they have to look amazing — of course.

You go back to Google and search for long-distance running shoes. Based on your search, you see a few shopping ads that are promoting exactly what you are looking for. You click one of the ads, which takes you back to one of the online stores you explored during your previous, generic Google search. 

Under the old UA model, the owner of this online store would see two distinct sessions: one from organic and one from a paid shopping ad. In GA4, though, it’s all the same session. 

The first event (or website visit) is attributed to organic search. All other events that follow the paid ad click are, then, attributed to the paid shopping ad. Similar to UA, but they are all categorized under the same session.

The midnight issue

This notorious issue arises from how UA logs its sessions. Say, for instance, that you are browsing at 5 minutes til midnight. In the back end, UA creates a row in its database for your sessions and logs all page views in that row.

Opposing this row in the database are columns identifying other metrics for the session, including one column for the date.

Once you cross the midnight threshold, UA needs to create a new row (meaning a new session) since it can’t have two different date values in a single row.

Now, in GA4, the date is connected to the event. Your session ID stays the same the entire time. Once you cross over midnight, the date and events change, but they are still associated with the same session ID. Again, this is enabled by the flat data structure in GA4.

In application, this means that you may see fewer sessions in GA4, but this will be more reflective of reality, since it eliminates the midnight double counting.

What happened to the UA metrics?

While the cleaner session tracking may be appreciated by web analysts and digital marketers, many may still have doubts about GA4. One of the biggest reasons for this skepticism is that many of our favorite metrics from UA have seemingly disappeared.

Ok. It’s true that some metrics may have “vanished,” it’s important to understand that Google is still tweaking GA4. So these metrics may re-appear in a new context or format as more updates are rolled out.

Take the bounce rate, for instance. I mean, “bounce rate” is so 2020. In its place, we now have the engagement rate. Think of engagement rate almost as the positive inverse of the bounce rate.

In UA, a bounce was triggered if a user entered your website and only one page view was registered before the user left.

Now, in GA4, an engaged session is registered when:

• A user visits two pages, OR
• A user triggers a conversion event, OR
• A user has the website or app open and in focus for more than 10 seconds

That last one is also used to calculate the new engagement time metric. Think back to our online shoe shopping example. Those 30 tabs would count as sessions in UA the second they were opened. Even if they are not the active tab in the browser window. Now, they don’t count as an engaged session until they are “font and center” for the user.

It makes sense, right? You might miss bounce rate, but it is arguably more important to focus on the users that are engaged on your website and which content is keeping them engaged.

Improved attribution

The Universal Analytics attribution model took major cues from the fact that the platform was so focused on sessions. It applied a last-click model, which provides total attribution to the last click (like an ad) before a session was initiated.

All other advertising and promotions that influenced a user before the click are ignored. While that may be great for performance marketers, it doesn’t accurately reflect the reality of your advertising ecosystem.

Thanks to GA4’s event-based model, you can apply any predefined data model to the platform. To be sure, the data-driven attribution model is still the default in GA4, but you now have options.

Google has kept the specifics of its new attribution options hazy (likely intentionally), but here’s what we do know about what it takes into account:

• Time to conversion
• Device type
• Number of ad interactions
• Order of ad interactions

Specifically that last item should help to provide a better perspective on the contribution of your different advertising tactics. While it’s nice to know which ad creative is driving the last click, it’s probably more helpful to understand how all of your ads drive conversion together.

Shifting to focus on what matters

We understand. It’s tough to get used to all of the new features and formats, but for the most part, GA4 is trying to shift our focus toward what really matters.

For instance, users are great, but digital marketers and web analysts really want to dig into the actions (i.e. the events) those users are taking on the website. They want to know that session counts are accurate to reality. They want a realistic understanding of how each piece of advertising contributes to conversion.

They want a real picture of what is happening on their websites.

Google Analytics 4’s flat data model will help us better gauge user behavior rather than website performance. The new approach also holds some major advantages for e-commerce sites. Plus, the flat structure will allow for powerful machine learning applications (though we’ll dig into this more in a later article).

With the sunset of UA fast approaching, it behooves you to start digging in and getting a feel for GA4. Just take your time; give yourself space to explore while understanding that Google is still tinkering with the system, too.

And who knows. You may start unlocking some incredibly powerful perspectives and insights yourself.