You probably have come across the news, received an email, or encountered a pop-up in Google Analytics regarding this change: standard Universal Analytics properties have stopped processing new hits since July 1, 2023. The deadline for 360 Universal Analytics properties has been extended to July 1, 2024.
If you have recently transitioned from Universal Analytics (UA) to Google Analytics 4 (GA4), you might have found yourself pondering over the question, "Why does my data differ between UA and GA4?". If this situation sounds familiar, we would like to guide you through six potential reasons that could shed light on the data discrepancies you are observing.
1. Key metrics
Key metrics like users, sessions, conversions, events, and more have different definitions and counting methodologies in UA and GA4. Let's take the example of users counting. UA properties use the Client ID, whereas GA4 properties use the User ID. On top of that, UA properties display the total number of users, while GA4 properties show active users.
Another metric that differs is sessions. In UA, sessions "break" at midnight, which means that if you start browsing a website at 23:59 and continue until the next day, UA would record two sessions. In contrast, GA4 does not restart sessions at midnight, so it would only register one session for the same browsing activity.
To gain a comprehensive understanding of the metrics differences between UA and GA4, I recommend referring to this article that provides a detailed comparison of all the metrics variations.
2. Reporting Identity
There are 3 ways of identifying users within GA4:
- Blended: this identity method takes into account User ID, Google Signals, Device ID, modeled data.
- The User ID is a unique identifier typically generated during login, based on the email address for example.
- If no User ID is collected, information from Google signals will be used. Google Signals is a feature providing cross-device data from website users who signed into their Google Account (e.g. Gmail) and who have turned on Ads Personalization.
- If neither User ID nor Google Signals is available, the Device ID will be used. Device ID, also known as Client ID, is a random identifier that is automatically generated for a device. It should not be confused with the User ID.
- If no identifier is available, then modeling will be used.
- Observed: this method considers User ID, Google Signals, Device ID. It is similar to the Blended identity approach but does not involve modeling.
- Device-based: this method relies solely on Device ID.
3. Conversions counting
In UA, a goal was counted only once per session. When GA4 introduced conversions, a conversion was counted every time the event fired. Since then, there has been a change in GA4, and it is now possible to select the counting method on an individual conversion level. This means that you have the flexibility to choose whether a conversion should be counted once per session (similar to UA) or every time the event fires.
It is important to note that this change in counting method is not applied retroactively. If you make changes to the counting method, the data will be counted differently for conversions that occurred before and after the change.
4. Modeling
Modeling refers to the use of machine learning techniques to fill in data gaps and estimate conversions that cannot be directly observed. Modeling is a feature exclusive to GA4. There are various examples of conversion modeling in GA4, some of which are highlighted in this article.
- One such modeling technique used by GA4 is behavioral modeling for consent mode. Consent mode is used to model conversions that are missing for users who have declined analytics cookies. It leverages the behavior of users who have accepted analytics cookies to estimate conversions for those who have declined.
- Additionally, Google has announced the use of enhanced conversions for modeling online conversions starting from Q2 2023. Enhanced conversions is a feature that enhances the accuracy of conversion measurement when cookies are unavailable, by leveraging first-party data.
5. Filters
In UA, filters are a powerful tool used to modify data within a view. They also allow you to include or exclude specific data from being processed. In GA4 properties, filters are currently not supported in the same way. The only type of filtering available in GA4 is the exclusion of developer or internal traffic and it is only applicable when you apply a filter.
6. Attribution
UA uses a last-click cross-channel attribution model while GA4 uses different attribution models based on the reports you look at. However, both UA and GA4 share a common practice of excluding direct visits from receiving attribution credit, unless the conversion path consists solely of direct visits.
If you recently set up GA4, it is possible to observe a higher portion of traffic attributed to Direct compared to your UA data. This is because GA4 has less historical data available to accurately categorize and attribute traffic on your website.
In conclusion, it is important to note that the list provided is not exhaustive, and there may be additional factors contributing to the data discrepancies between UA and GA4 properties. Achieving perfect data is unlikely, and some level of discrepancy is normal. However, if you notice a significant disparity between the two platforms, it is worth investigating to ensure there are no underlying tracking issues. If you would like assistance in delving deeper into your data, do not hesitate to reach out to us.
Justine Heeren
| LinkedinThis email address is being protected from spambots. You need JavaScript enabled to view it.