In October 2020, Google introduced the latest upgrade to its Google Analytics analytics tool, Google Analytics 4 (also GA4, GA4 property). This new version of the analytics tool completely changes the way we look at web analytics – all analytics in Google Analytics 4 is built on events (events + parameters). In Universal Analytics (now called GA3), events were used only for one report, and working with them was overly complicated.
Events are replacing all previously known interaction types from the original Analytics, such as Pageview (now page_view event), Transaction (now purchase event) and others.
Chaos and setting order in measurement
Thanks to Google’s quite aggressive communication that began in early 2021, there has been a bit of chaos in the marketing market.
It is a confusing situation for new users when only the GA4 property is created by default for new sites, so 90% of the tutorials on the internet do not work, and it is necessary to click on other settings and force the creation of the GA3 property.
First of all, we want to reassure you – you won’t lose your data in Universal Analytics (the current version of GA) just yet, and you certainly don’t need to set up GA4 as quickly as possible. The exact timeline is not set, but experts assume about a year of GA3 operation, eventually it is possible that it will be impossible to set up a GA3 property.
And certainly the main motivation for switching to GA4 shouldn’t just be that GA4 is the talk of the town. But it is an interesting incentive.
What is changing at GA4 Property?
A) Data model
It is important to note that in the case of GA4, this is not a minor change, but a transition to a completely different data model. The existing Universal Analytics are session-based analytics, and it evaluates the website primarily from the perspective of individual visits.
GA4 uses an event-based analytical model where the key is to identify individual users and then their endless stream of events and transactions. This model is familiar to analysts from Google Analytics for Firebase or GA4’s predecessor, Web+App.
B) Device perception
One of the marketing ploys of GA4 is that it should be possible to measure web apps and mobile apps in one place. The tool can indeed do this, but the setup is not entirely straightforward. What makes it even more complicated is when web apps and mobile apps work with completely different user databases. We can set up such a measurement, but let’s face it, how many companies will actually use this?
C) New reports (currently fewer reports)
Update August 2021: Google has released a new version of GA4 reporting that allows complete customization of the left menu and all reports. These settings are called Collections, and so far it is possible to create so-called Overview Reports, created with tabs and widgets, and detailed reports created similarly to GA3 reports, with a graph at the top and a table below.
Due to the overall change in the data model of Google Analytics, the reporting must also be modified. If you already measure with GA4, you may find that there is not much there when you open the interface. And you’re right.
Most of the reports that you know from your current measurement will not be there. So you are left to report on your own either via Analysis Hub (where there are perfect funnels by the way) or via BigQuery + e.g. Google Data Studio, which can be a big obstacle for less advanced projects.
D) New technological possibilities and perspective
Update August 2021: We have already had a dozen GA4 implementations with our clients, and it turns out that the new event-based data model, which you can name by yourself, provides a lot of freedom in the design of the event structure. If you choose the right naming system, you can create dozens of events that are obvious at a glance what they are for and how to work with them. At the same time, you will appreciate the possibility to modify and create events directly in the GA4 interface, especially for more complex implementations, where 1 event leaves GTM and is further sorted according to parameters in the GA4 interface.
The unification of the data model to events only opens up better possibilities for machine learning models (e.g. anomaly detection), but at the same time it places greater demands on the technological maturity of the company and the technical design of the measurement.
Should I switch to GA4 then?
Are you asking if and why to fully switch to GA4? That’s what we asked too. That’s why we’ve created a simple balance sheet of PROs and CONs that will surely evolve over time. It is now (21.1.2021) evident that the arguments against are prevailing for the time being. We anticipate that the pros will increase, and when the cons outweigh the cons = there will be widespread adoption of the new measurement across the marketing market. What are the advantages and disadvantages of adopting GA4 now?
Update August 2021: What has changed in more than six months? How do we view the issue now? Our recommendation to implement GA4 in parallel is still valid, and we recommend doing it as soon as possible. There are two main reasons for this.
Why switch to GA4?
The main recommendation from Google is to measure in parallel – i.e. in classic Universal Analytics and in GA4. We also lean towards this with the motto: “Those who are prepared are not surprised.”
A modern approach for new projects
If you are currently starting a new project, a new web application, portal or blog platform, it is not worth starting with GA3, but we recommend starting with GA4. For one thing, you can get a head start on your competitors who will find it difficult to get off track, and you won’t drown your time in building duplicate processes for GA3 and GA4 reporting.
Availability of historical data
If you set up GA4 measurement today and also enable GA4 export to Big Query, we will have a time series of data that we can use for yearly and monthly comparisons. The sooner you start collecting the data, the sooner you will be able to fully transition to GA4 reporting in the future because you will have existing data. So if I start collecting GA4 data today, I will be able to do the first YonY (year-on-year comparison) in 2 years. Alternatively, in 1 year and 1 month if we are talking about comparing monthly data for the previous period. Part of this is related to the next point.
New baseline for legacy data
It is likely that the baseline for the metrics that you have been used to will partially change when GA4 is implemented. For one thing, they are counted differently, some have disappeared for now (August 2021) from the reporting interface (e.g., depth of visit, pages per visit) or they have been replaced by a different concept (e.g., bounce rate has moved to engagement rate; sessions are still counted but they stand in the background.)
So it’s best to start collecting data earlier and start getting used to the new metrics and associated averages that your project generates. Even switching to the concept of users can be a big leap for some projects that primarily report visits (sessions).
Funnels
The Analysis Hub (renamed to Explore > Explorations in August 2021) currently has a very well-developed funnel interface. It is one of the best solutions we know of. Plus, it avoids the complicated bending of Events in current Google Analytics into a form to make it meaningful.
More benevolent data sampling
According to initial information, GA4 has a more benevolent sampling, similar to Google Analytics for Firebase – i.e. it returns unsampled data even with larger data.
Update August 2021: Standard reports (unadjusted) always return aggregated data from 100% of the data. If you create custom reports (called Exploration), sampling will appear when queried for more than 10 million events (partially analogous to the 10 million hits in GA3). Similarly, you may occasionally encounter so-called Tresholding, which is also related to privacy and the ability to trace specific users in reports. “Unknown” records may appear in the rows.
The processing time of data into standard reports is 48 hours from the reception of data. However, it will often be faster.
Event Sets
Within GA4, there are sets of recommended events and automatic events that are partially self-measuring – a benefit for some, not for others. However, for the first time ever, Google is providing lists of recommended events that should be measured on the site – these are worth sticking to and we believe they will evolve quite a bit over time.
In practice, we have found that it is appropriate to use these events as inspiration, but not all events have a fully descriptive title. In our implementations, we often work with a system of prefixes and suffixes that form a unified event naming system. For example, chat_open, chat_message, chat_close, etc.
Why not switch to GA4 + GA4 news
The list was updated in August 2021
- Google Analytics Universal Analytics does not yet have an end of support date = no need to switch hastily and at any cost (due to the technology-intensive maintenance of GA3 for Google, it will probably be sooner rather than later)
- In GA4 the visualization of outputs is still difficult (the easiest way is BigQuery > Google Data Studio) – you will struggle with nested fields, the GDS connector for GA4 has undergone some changes, but when pulling parameters from events you are still referred to custom dimensions, of which there are only 50, this situation will probably not change much in the free version
- A different philosophical approach, to which the whole company and other connected parties need to gradually adapt – this point will always be relevant and it’s up to you when you start
- GA4 can currently be fully integrated with Google Ads and start using them as a source of conversions and audiences, which are great by the way
- Lack of the ability to create Views – so the whole data is always shared and if you want to share only part of the data, you have to go the way of custom reports (even individual Streams are not yet shareable) – this is not yet solved, currently you still have to share all the data
- Missing reports that you are used to – in August 2021 the possibility to create custom menu structure and reports via Collections was added, but it is still not possible to replicate all reports from GA3 without using external tools
- Data filtering on output (currently data filtering is replaced by event modification)
- Fewer prepared metrics – calculation of most metrics is delegated to you as a tool user
- Need for more company involvement in Google Cloud (BigQuery), Google Data Studio
- Lack of Best Practices – so far only a “toy” for advanced analysts, technology innovators and large companies; however, the situation is gradually changing, the GA4 interface is still evolving and even new articles on GA4 are quickly becoming outdated
- There are already enough documented GA4 features, but still no explanation of all dimensions and metrics can be found
- Even for simpler measurements you need to do an initial study – the technical design of the measurement is much more complex compared to Universal Analytics (in UA you generally worked with Event Category, Action and Label + 20 dimensions) – here it depends very much on the maturity of the project (if you are used to working primarily with, for example, Facebook Pixel, the transition will be easier for you, because the data concept is fundamentally the same)
Update August 2021: Our current opinion, based on our experience with dozens of clients, is as follows: Understanding GA4 is challenging if you’ve never worked with them. The frequent changes to the reporting interface and settings put pressure on users to constantly monitor changes to the tool and live in some uncertainty as to whether the tool will work the same tomorrow as it does today.
Based on discussions among analysts and marketers, it currently looks like GA4 is putting pressure on analysts to actually work with them. The logic that can be seen is that there is a push to differentiate the roles of analyst and marketer.
The analyst should ensure the collection of quality data and then create reports – either in the Explorations section, using custom reports in Collections, or using external tools such as Google Data Studio, which will then be used by the specialist.
Even with the fact that anyone can create the event names they want, we can assume that the days of giving a specialist View and Collaborate rights to Property and letting them find what they need may be over. This may not be the case, but it certainly won’t be without perfectly maintained documentation.
As a result, discussions about GA4 adoption often end up on capacity and time issues with already overwhelmed and often understaffed marketing/analytics teams. They simply don’t have the right cup for a project that numbers in the tens to hundreds of hours.
We believe the situation will change within the next fiscal year, where the line item “transition to GA4 – 100 – 200 hours (external + internal work)” should not be missing from your marketing budgets. And if you can do it faster, you can do some well-deserved team-building with the savings. Note: We’re talking about complex projects that require outsourcing analytics services, as we do for our clients. If your website runs on any CMS platform, “switching to GA4” is a matter of an afternoon for you.
The main reasons AGAINST the transition are the “leaky” documentation and the need for a complete change in access to data from the web, which requires training of employees. However, the need for a change in philosophy and staff training can never be avoided and we are happy to help you with this point. The third major con is also the process changes needed, for example in the areas of visualization and reporting.
As far as more advanced visualization and reporting is concerned, a necessary intermediate step is to set up data export from GA4 to BigQuery. This can of course be done with one click and by setting a credit card, but it is recommended to familiarize yourself with at least the basics of Google Cloud beforehand, because to export data you need to set up a Google Cloud project with everything that goes with it (security, data processing, access rights, payment methods, etc.). We do not want to scare you, it is possible that the free version (sandbox or free-tier, which Google has for its services) will be enough for you, but in fact, you are adapting a new technology into your company and the associated legal obligations (at least DPA data-processing agreement, and others). The easiest way to create a visualization from the exported data is to use Google Data Studio.
So how to adapt GA4?
I have a long-running eshop
If you have a working e-shop, where you have Enhanced E-commerce set up and everything works, you use Google Analytics audiences for campaigns and report everything at the Session level, there is no need to frantically switch to the new GA4. You can be completely at ease.
We recommend creating a duplicate measurements in the new GA4 data model (i.e., create a new GA4 property under the same account), become familiar with everything, and use the existing data layer to duplicate the event in the new measurement.
If you are not using a data layer, we recommend that you consider implementing one in your current technical site design. This will refine your measurement and give you more control over it.
At the same time, you can train your employees and familiarize them with the new GA4 interface under the supervision of an experienced analyst. As part of employee training, we recommend you also focus on process changes and consider areas where you will prioritize the transition to the new model.
I’m planning a redesign
If you plan to launch a new version of the website in the next six months (major redesign, foreign mutations, etc.) we recommend implementing the data layer in the new version of the website. Along with the implementation of the data layer, we recommend running both GA3 and GA4 measurements. We definitely do not recommend implementing hard-coded versions of the GA3 and GA4 libraries directly into the site, rather choose Google Tag Manager. GA3 is worth setting up if you want to work with a historical data baseline from a previous project, or another language version. If this is not relevant for you, start with a clean slate in GA4 straight away.
If you are interested in the functioning of GA4, the functioning of data collection into GA4, the collection limits and their blocking on the user side, you can set up both versions of measurement and make regular comparisons in the technical audit.
I’m starting a new project
If you are starting a new project, we recommend that you start straight away with GA4. This will save you a lot of time setting up duplicate measurements, and you can be sure that all your reporting will be future-ready. You can use the time gained to keep up to date with the latest news around GA4, allowing you to quickly adopt new features.
I am starting a new project and want to be proactive and future-oriented
Are you starting a new web project in 2021 and beyond? Do you care about the security of your users’ data and the protection of their privacy? Are you concerned about user consent related to the ePrivacy Directive, which will tighten up what and how we can collect data and use it in campaigns?
If you answered yes, you should focus on two more points in your GA4 implementation: the Consent Mode and Server Side Tagging. These two components of modern web analytics allow you to have complete control over what data you collect about users, in what form, and also allow you to get everything ethically and legally correct. You will also get more data and refine your management decisions.
Want to find out more? Contact our professionals.
Future outlook with GA4
Update August 2021: The following has turned out to be true. Google has indeed kept their word and the development of GA4 and especially their interface is moving forward. Occasionally there are some bugs or unexpected “features”, but in general we can say the tool is moving forward.
Due to the complex adoption, they have suffered especially on Twitter or in the Google Analytics 4 analytics community not only hate, but also from big names in SEO or PPC campaigns. However, we are optimistic and believe that the frustration of companies and specialists comes mainly from the lack of time they can devote to GA4. We therefore recommend considering outsourcing analytics services to professionals.
As time goes on, there will undoubtedly be new benefits and reasons to move to the new GA4 measurement and associated new processes.
Currently, our clients are finding the greatest benefit in measuring complex funnels, calculators or advisors where you can take full advantage of the capabilities and new funnel features directly in the Google Analytics interface.
We recommend you avoid the tempting “go to GA4” buttons that jump out at you within the Google Analytics interface for now. As you’ve read above, the change you’d embark on is more complex than it first appears.
If you are considering switching (or duplicating measurements) to GA4 and need help with an initial consultation, or a detailed discussion of any of the above, please drop us a line.
