Google Analytics 4: Are you ready to implement it?

OCTOBER 2020 / BY FLORIAN PERTYNSKI

This article hopes to help analysts and users of Universal Analytics who are considering if and when to implement Google Analytics 4. For the past year, we have implemented and supported App+Web for a number of clients, picking up valuable lessons along the way. Here is our assessment of the features, functional status and notable comparisons of GA4 to Universal Analytics in the context of the most common use cases, as well as the rationale as to why it is important to adopt as soon as possible.

_
App + Web vs. GA4: what’s changed and why it matters

On the 14th of October, 2020 Google Analytics 4 was announced as Google’s new default analytics platform for websites. This adventure, however, started in 2019, when Firebase Analytics for mobile apps started supporting the GTAG framework, thus giving birth to an integrated mobile app and website measurement platform: App+Web. The product was groundbreaking, but we could see organisations using Universal Analytics were rather cautious to adopt it. The reason? Firstly, there is a different data format, with log-style events instead of sessions, pageviews and interaction tracking events. In addition, the Beta version was missing popular features of Universal Analytics, which users were accustomed to. To make up for the missing features, the months following the release saw numerous improvement releases, bringing App+Web functionality coverage closer to Universal Analytics.  Now, 14 months later, with App+Web stepping out of beta as Google Analytics 4, we can say it is a completely different product, feature-rich and business focused. Are you ready to implement it?

_
Why implement now

You’d better be, since phasing out of Universal Analytics is a question of time (years rather than months, though). So before the sun sets on Universal Analytics, let’s consider why we should all implement GA now.

  • Back history of data
    • The earlier you implement GA4, the more history in the new data format you will have to work with.
    • Think of the depreciation of GA for mobile apps earlier this year, when the data sets were removed. So those who had switched to Firebase/App+Web in time, managed to save some historical app data. We don’t know how Google is going to deprecate Universal Analytics, therefore, it is best to stay on the safe side.
  • Time for learning
    • GA4 is different from UA.
    • You will need to learn to think in terms of events and parameters, learn to use the interface, learn to build custom analyses – this all requires time. 
    • The cherry on top: you will find great benefits by brushing up on your SQL for BigQuery.
  • Organisational buy-in
    • There are new data concepts and definitions that your organization will need to learn and feel comfortable with.
    • People need time to adjust to new ways – ensure there is enough time to build confidence and trust.
    • We recommend running Universal Analytics and GA4 in parallel. Before the management and your colleagues accept and get used to the new data concepts, they may want to rely on Universal Analytics for a while.

_
Is GA4 ready to implement now?

To answer this question, let’s consider the common use cases of Universal Analytics, which – if unavailable – are likely to cause frustration of GA users and confusion in their organisations. In other words: is GA4 ready to meet your expectations and business requirements?

GA4 stems from event-based mobile app analytics therefore is different from the session-based Universal Analytics. Since the launch of beta last year, however, the GA4 product team has been working hard to make sure that the common Universal Analytics use cases are preserved.

_
Expectations and challenges

Stemming from event-based mobile app analytics, GA4 is different from the session-based Universal Analytics. Since the launch of beta last year, however, the GA4 product team have been working hard to make sure that the common Universal Analytics use cases are preserved.

1. Session reporting

This was added to GA4 at an early stage. The available metrics are, among others:

  • Sessions (count of)
  • Engagement rate (more or less the opposite of UA’s Bounce Rate )
  • Engaged sessions
  • Avg. Engagement Time per Session (more or less, time on page with browser tab in the foreground)

2. Ecommerce

Ecommerce was included in the roadmap since the launch of App + Web. For a few months the data collection methods for measuring ecommerce were already available, but no Ecommerce reports in the interface. It changed with the announcement of GA4, with full-fledged ecommerce reporting available . The only missing elements are:

  • funnel reporting in the interface
  • ecommerce support in Google Tag Manager GA4 tags

But I do not doubt Google are working on it.

3. Rollup reporting

Sending data to multiple properties – such as individual site and rollup property – is available for websites only . With mobile apps, you can send the data to a single property only. Not sure if this is going to change any time soon due to Firebase architecture.

4. Google Ads integration

Full integration allows to easily create Audiences, covering some extra bases compared to Universal Analytics:

  • a quicker setup – with GA4, Audiences have been merged with UA’s Segments, so setting up an Audience is simply fewer clicks and less typing
  • time-based conditions in sequential audiences
  • multiple audience templates:

Fig. 2: GA4’s multiple audience templates

5. Conversion reporting

For Analytics users who rely on conversion rates (e.g. newsletter subscription rate, contact form submission rate etc.), this is where right now GA4 does not tick all the boxes. Right now, you can mark interactions (events) as Conversions in order to use them in Google Ads. In GA4, it is impossible to report on Conversions one particular event – the metric you get is aggregate of all Conversions. 

(For those with SQL skill or resources, this poor conversion reporting should not be an issue since it is possible to calculate conversion rates in BigQuery. Still, it is an additional hurdle.)

6. DataStudio integration

A recent addition to GA4, the feature is incomplete now. Currently, these are the chief limitations:

  • Event parameters cannot be used in the dashboard – only events.
  • Audiences are also unavailable, so you rely on DataStudio filters.

(Again, BigQuery as a data source may come to the rescue, although preparation of the data will require SQL.)

_
New opportunities

While some may still be uncomfortable with the new data format and the few missing features, GA4 opens up some new areas with great business and insight potential

First and foremost:

1. BigQuery integration

Universal Analytics required the premium tier licence to activate this integration, but with GA4 it is a standard feature. BigQuery itself comes with storage and usage fees – while these will be negligible for most organisations, calculate your costs before activating the integration.

For those new to BigQuery – in a nutshell, it is:

  • It is a fully managed, scalable cloud data warehouse.
  • It comes with an on-demand pricing model (i.e. you pay for the amount of data stored and accessed).
  • BigQuery is a native data source for DataStudio, Power BI, Looker, Domo and other data visualisation tools.
  • You can think of BigQuery integration as a first step in a Machine Learning project, with BigQuery ML available out of the box – for starters.

2. Cross-platform user identification

In Universal Analytics, User ID can be a basis for user-centred cross-platform analysis. It is available in User-ID enabled Views and gives insight into device overlap, device paths toward conversion etc.

GA4 takes user cross-platform user identification well beyond that by making user ID a default user identifier. Thus, 

  • User ID is a basis for the Users metric, which results in a more reliable user count.
  • This way, User Properties are shared across devices. E.g. if you assign a (signed-in) user a user-scoped attribute, that attribute will apply if the user visits your site (or app) from a different device.
  • It allows cross-device targeting of users in Google Ads, giving your campaigns greater reach within a qualified audience.

3. Machine learning

GA4 uses machine learning to deliver insights:

Fig. 4: ML in action: anomaly detection in reports using a line chart

It also uses predictions to build Google-Ad-ready audiences. This is so advanced! Combining the power of Machine Learning with the reach of Google Ads makes Google Analytics 4 a marketing powerhouse:

Fig. 5: Audiences built on predicted metric values!

I’m sure that GA4 will make use of ML more extensively. In their blog post, Google show a churn probability report. In the GA4 properties which I have access to, I unfortunately cannot see the metric. All the blog posts about GA4 also show the same screenshot, so I assume others can’t see it either 🙂 As we have often seen, however, Google release product features gradually, so let’s wait and see.

4. Sampling

Initially we were overjoyed by the fact that GA4’s predecessor Firebase Analytics didn’t use sampled data in reports. With GA4 it is partly true: 

  • the out-of-the-box reports are not sampled,
  • but the custom reports (in the Analysis section) do not have that privilege.I haven’t seen documentation regarding GA4’s sampling thresholds, but they could be similar to those of Universal Analytics.

If your reports are heavily sampled, data export to BigQuery can be a solution. In that scenario, you can access 100% of your data. You can also use your BigQuery dataset as an unsampled data source of your visualisations (in DataStudio, Power BI, etc.).

_
Summary & closing thoughts

Google Analytics 4 is constantly developing, with fewer and fewer shortcomings versus Universal Analytics. The new opportunities with BigQuery integration, data quality and machine learning clearly offer advantages. Here is a head to head comparison of Universal Analytics versus GA4 in terms of the primary features:

With little to lose and valuable experience to gain, we hope this article has provided you with clear arguments to justify starting your implementation as soon as possible:

  • Gain and preserve historical data – the sooner you start, the more data you will have.
  • Start your learning curve now and gain a competitive edge with the benefits.
  • Stakeholders need time to adapt to new ways – be responsible and start early.

As always – let us know if you have any feedback and drop me a line at florian@iihnordic.com if we can help you further.

Florian Pertynski

Analyst

florian@iihnordic.com
+45 65 74 66 62

Florian Pertynski
Analyst
florian@iihnordic.com
+45 65 74 66 62