5 Key differences between Universal Analytics and GA4

We are now 5 months away from the deprecation of free Universal Analytics (UA) properties on 1 July 2023. If you want to continue using Google Analytics (GA) you have to set up Google Analytics 4 (GA4). Surprisingly, many organisations still have not done this, or have done so at a very basic level. Basic is better than not at all, but it would be wrong to treat 1 July as a switch-over date rather than making the investment to set up GA4 well as soon as possible. The reasons for this are:

  1. You will want as much historical data as possible in GA4 so that you can easily make comparisons to previous date ranges in GA4.
  2. You will want time to iron out any tracking or configuration issues before the 1st of July deadline.
  3. You need practice to get to grips with GA4. It’s a very different tool to Universal Analytics and there are many different concepts to get your head around.

It’s this last point that we want to start to address in this blog post. I say “start”, because the differences are so numerous that this could potentially turn into a book if we tried to list them all. So this post will try to focus on the key differences at a high level. However, please note that even this could be called contentious. Ultimately, a feature that’s important to one organisation may not be important to another, so please give me a bit of latitude on this.

So here goes. This is what DataTribe sees as the key differences between UA and GA4 and why it’s important.

Unified mobile app and website tracking

If your organisation has mobile apps as well as websites you can now track both in the same GA property. With Universal Analytics you would normally have to track websites and apps in different properties so that the data was completely separate.

Why this is cool?

It means that you can now analyse the engagement of your users in one place. This is especially useful if you have functionality or content that is both in your website and mobile apps. For example, for TVNZ or Newshub, the same news article would appear on both their website and their mobile app. Being able to see how many users viewed content and how this was divided across app and web without having to first extract the data is incredibly useful for those organisations. 

Another advantage is that you can understand how individual users (see my next key difference might be interacting with both your mobile app and your website.  This is important in situations where a user journey might start on the website but continued on a mobile app, or vice-versa.

What are the limitations?

You’d need to ensure that you had a way to track cross-platform content with the same naming convention across both your website and mobile app. Because websites have URLs and Page Titles, while mobile apps have Screen names and Screen classes.  In the example of the news sites, you’d need to also build in tracking of the article name so that you were able to do the cross-platform analysis described.

For cross-platform user journeys, you’d need a way to identify the user across both platforms. Usually, this means they have to be logged in to both the website and mobile app (although see also Google Signals in the following section).

Unification of user data across devices and platforms

If a user is logged in or has been identified through Google Signals across different devices (desktop/mobile) and platforms (web/app) then they will be counted as the same user across the devices and platforms. With Universal Analytics users would be identified through a 1st party cookie so if you had visited a site from different devices/browsers then you would have a different cookie on each device and would be counted as multiple users.

Why this is cool?

More accurate user counts than in Universal Analytics. Ability to analyse how users engage with your digital content across different devices and platforms, and see the level of cross-over for each.

What are the limitations?

As described, the user will only be unified if they are logged in using the User ID feature, or if Google Signals is enabled (at a very basic level, Google Signals basically means that the user is logged into their Google account on their device/browser and have ad personalisation enabled). If neither of these is applied, then GA4 can only identify the user through a 1st party cookie or app instance ID, which won’t match across devices or platforms.

If you enable Google Signals in your GA4 property, while this does give you a heap of useful audience remarketing features, it also means that your reports in GA4 may have some missing data due to Data Thresholds.

Different data models

Universal Analytics collected data as “hits”, which could be “events”, “pageviews”, “e-commerce” and a few other hit types. Events would also be tracked in 3 dimensions: Event Category, Event Action and Event Label.

By contrast, all data collected in GA4 is an “event”. A page view is a type of event, just as a download is a type of event. When you send an event to GA4 you can also add parameters. Parameters are like the metadata of events. They give context to the event. You can give those parameters any name you want so that it’s more relevant to your organisation and the specific event being tracked. For example, if you send a “video_start” event, you can also send contextual data like “video title”, “video length”, “video URL”, “video type”, “video platform”, etc. Essentially, anything that is going to be useful for you when it comes to understanding how your website is performing and what users are engaging with what content. You are not restricted to the non-descriptive dimensions of “Event Category”, “Event Action”, “Event Label”, as field names under which to send event data. In fact, you can send up to 25 parameters with each event with GA4.

Why this is cool?

It allows you to customise the data you collect much more than you could with Universal Analytics. Custom data means relevant data. Relevant data is more valuable data.  Reporting, analysis and building remarketing audiences (or activation if you prefer that term).

The ability to send up to 25 parameters allows you to capture much richer data, which leads to greater insights and remarketing enablement.

What are the limitations?

The GA4 data model focuses on Event and Session data, and less on Session data. Universal Analytics was more focused on session data with session-level dimensions and metrics. While you can customise the parameters you send with GA4, at the time of writing you cannot create Session scoped custom dimensions or metrics. We are told by Google this is coming, but we have also been waiting a long time.

No sampling (kind of)

Hands down, the worst thing about Universal Analytics is sampling.  This is where you run a report over a specific date range and UA would return results based on a sample of sessions.  The longer the date range or the more complex your report, the more likely that it would be sampled.  Depending on the date range and the complexity,  if the data was sampled, the data returned could be based on a sample of anything between 1% to 99%. The lower the sample size the less reliable the data becomes if you want to take any action on it.

GA4’s standard reports are largely unsampled (Yay!). There are conditions where if you run a report that covers a very high number of events, the reports will be sampled, but in general, there is a much reduced likelihood of sampling.

GA4 also offers the ability to copy all the raw GA4 data to BigQuery, an analytics database, which will never be sampled and allows you to analyse the data and work with it in any way you want.

Why this is cool?

Without sampling, you can feel much more confident in the accuracy of your data and you can take action with more confidence.  The better the quality of the data you make decisions from, the better the business outcomes of your decisions are likely to be.

Being able to have all the raw event data in a BigQuery database is really cool too. For those organisations that have the technical resources to query this data, the possibilities for analysis and joining other data sources (both internal and external) can be invaluable.

What are the limitations?

It’s not so much that sampling under GA4 has many limitations, as there are additional ways in which GA4 data is limited in ways that weren’t done with Universal Analytics.  Some examples are:

  • Data thresholding which leads to missing data when you have Google Signals enabled  when the data doesn’t meet the  threshold (see earlier).:
  • The “Explore” section reports (the custom report building section) is limited to the last 14 months of data.  This means you can’t create year-on-year reports in the Explore section.
  • The GA4 API has quota limits.  The GA4 connector in Looker Studio (formerly Data Studio) uses this API.  The quotas are not particularly generous and many organisations have found themselves reaching the hourly or daily limits very quickly.

Also, with respect to BigQuery, you need the technical resources to be able to leverage this, and, depending on volumes, there can also be storage and querying costs involved.

The interface

Because GA4 is a completely different analytics platform based on a different data model, rather than an evolution of Universal Analytics, the interface is completely different. The main reporting sections of the GA4 interface are:

  • The standard reports section (where you can add your own custom reports to a limited extent).
  • The Explore section is where you can build all manner of different custom reports and explore individual user journeys.

Why this is cool?

The Explore section is a fantastic tool.  It allows you to build a number of different report types:

  • Free form (similar to reporting capabilities if you were using Looker Studio)
  • Funnel
  • Path exploration
  • Segment overlap
  • User Explorer
  • Cohort exploration

What each of these does could actually be divided into 6 different blog posts. However, suffice it to say that it leaves the custom reporting section of Universal Analytics in the dust.  It is awesome6T.

What are the limitations?

The Standard reports section is good, but not as good as the Universal Analytics standard reports section (in my opinion). The reports are organized well but there is a more limited range of data. While admin users can build custom reports in this section, I’ve found that it doesn’t allow me to build many of the reports I wanted to build. 

Also, as mentioned earlier, the Explore section is limited to the last 14 months of data for free accounts.

Final thoughts

There are so many other differences that could have been included in this blog post. To list some of them:

  • More ability to report using different attribution models (UA was mainly the last click).
  • Better audience creation capabilities
  • Worse Google Ads reporting
  • Better user privacy features
  • No ability to create views under each property
  • A different way of tracking goals

Some of these might have been more relevant to some people depending on the context. This post simply gives an indication of the priority of changes that we believe impact users the most.  

As you can see there is a lot to take in when it comes to changes. This is why we recommend that you implement GA4 well, as soon as possible.  It will give you time to practice and explore the tool so that when you are no longer able to use Universal Analytics, you don’t feel lost. Implementing GA4 well does not mean simply replicating what you were tracking in Universal Analytics. It means leveraging what GA4 is designed to do. After all, it’s a different tool.

If you need an extra hand with GA4, the team at Data Tribe will be happy to support you. We can support you with a package that includes:

  • Creating a measurement strategy and tracking plan
  • Setting up GA4
  • Creating dashboards and custom reports from your GA4 data
  • GA4 training

Contact Nick at [email protected] to learn more about our GA4 Implementation package. 

And if you’d like to read more about GA4, don’t forget to visit our blog. We’ve already covered some topics: