The entire marketing and analytics market has been hit by news announcing the official end date for data processing in Google Analytics 3, known mainly as Universal Analytics. Some were positively affected by this news, some negatively and some may have been terrified. For that reason, we have prepared this article to set the record straight on what lies behind this announcement. And, most importantly, to advise what managerial solution is currently on the table.
As of mid-March 2022, you’ve probably started seeing this message in your Google Analytics accounts. This is the next step in Google’s roadmap to convert all users from GA3 to GA4. We regularly update our article about the GA4 situation here.
What is behind the above message:
- New data can be collected in Universal Analytics until July 1, 2023 (until October 1, 2023 for paid GA360)
- As of July 1, 2023, historical data should still be available for approximately 6 months
- Google recommends exporting this data, for non-paid accounts you have to unfortunately go the tortuous route of exporting via Sheets, CSV or Excel, or choose the not-so-kosher method of daily export via API – for example using Keboola
- The final date for the complete end of GA3 will be announced in the future
- If you are currently using Google Analytics, you are probably using the GA3 version for accounts created before October 14, 2020, when GA4 gradually became the default option
- Google itself already recommends switching to GA4
- If you’re using GA3 to import conversions and audiences to Google Ads, we recommend starting the transition to the GA4 option
What steps can you take today?
Instruct your analyst or marketing team to ensure immediate implementation of GA4. If you want to have at least a year of data for year-over-year comparisons at the time of the forced transition (July 1, 2023), you now have approximately 3 months (March 21, 2022) to complete this implementation. According to our experience, depending on the complexity of the e-shop, the complete transition can be achieved in 8 – 16 weeks, depending on the client’s cooperation.
Prepare an internal GA4 adoption schedule including deadlines. This may include the following steps:
- Setting up the business case for the whole project
- Audit of GA3 implementation
- Account structure design – property, data streams
- Setting up accounts and property
- Creation of data streams
- Setting up data processing conditions (User-ID, Google Signals, retention time, connection to Google Ads, connection to Search Console)
- Migration table for GA3 > GA4 events
- Migration table for GA3 > GA4 custom dimensions and metrics
- Migration table for GA3 > GA4 goals
- GA4 implementation itself (we 100% recommend Google Tag Manager)
- Checking, testing and comparison of GA3 vs GA4 data – discovering the new baseline
- Optional: Migration of audiences and use of predictive models
- Optional: GA4 integration + CRM using a data platform such as Keboola or Google Cloud
- Optional: Moving reporting from GA3 data to GA4 data
- Sharing GA4 accounts and reports across the company
- Series of training sessions across the company
Start preparing your developers to implement the data layer for existing projects (if the planned lifecycle is 3 years or more), or alternatively, integrate the data layer into the projects currently under development, ideally in the GA4 recommendation. The proper implementation of the data layer is a topic in its own right and is covered in a separate article.
What changes to expect from a business perspective?
GA4 may still change significantly over the coming months. The reports and the customization options offered by the tool may change as well.
It is possible to expect that the limits for free use will be adjusted or some features will be charged.
The baseline for key indicators will change:
- Users – GA3 and GA4 have different approaches to user identification, GA4 uses a significantly more sophisticated system for device deduplication
- Visits – the concept of visits is still retained in the GA4 background but is not considered a key metric, so we recommend starting to move away from the logic of visits
- Conversions – GA4 no longer uses the concept of goals. By making the entire measurement event-based, events are considered as conversions; it is also no longer the case that there can only be one goal met per visit, each event is handled uniquely. Therefore it is clear that GA3 goals will not equal GA4 conversions.
- Ecommerce – GA4 doesn’t yet offer as comprehensive reporting for Ecommerce as GA3 but they are collecting the data (so it’s more a matter of time)
- Measurement customization – not all measurement customization features are yet available in the GA4 interface – e.g., channel grouping or content grouping cannot yet be influenced beyond basic rules
It’s crucial to be clear about what role Google Analytics web data actually plays in your business:
- how you work with it and what you do based on it,
- whether you’re actually getting value from the data,
- whether you’re collecting it properly and whether it’s being overestimated.
Accordingly, it is possible to set up a customized migration for you. The fact is that Google Analytics 3, as a free tool in which you don’t even own the data collected, has often become a crucial part of how companies operate online. Therefore it is self-evident to consider potential alternatives, whether it’s a completely custom solution or, for example, hosting Matomo Analytics on your own cloud instances.
Don’t panic! It’s clear that new and different means scary and bad for some people. I assure you that GA4 is fully future-ready and is adding a lot of missing features that are now very quickly reaching users of the free version. And learning something new is challenging but beautiful, isn’t it?
Need a consultation?
If the above makes your head spin, you probably need a consultation from an analyst. Feel free to contact our team of professionals who can handle the migration of your project from GA3 to GA4.
The complexity of the project is logically derived from the complexity of the actual implementation and from what you want to transfer. The extent to which the data layer is prepared is also a key factor. Thus, migration projects can range from units to tens of man-days.
