Sessions (or maybe better known as traffic) should not be anywhere near a measurement framework. Sessions does not positively correlate to whatever the business north star metric marketing is judged against.
In the world of Google Analytics, we think of sessions is an important metric with Google moving the goal posts from GA3 to GA4 it has also shifted how sessions are tracked. There is no like for like comparison for sessions in GA3 v GA4. For many Google Analytics (GA3) had become (wrongly) the source of truth and now with the shift to GA4 the event driven model has caused havoc in being able to understand channel performance.
The evolution of sessions
It makes sense how Google has evolved the tracking of sessions. With GA3 it was all about visiting a website via a web browser, with GA4 it could be visiting a mobile app or a website via a smartphone or connected TV.
The key differences:
GA3:
- There had to be a pageview load for a session to track.
- A session is only attributed to one campaign parameter.
- A new session starts with a new campaign parameter.
- A new session is started at midnight.
GA4:
- A pageview is not needed for a session to track.
- A session can have multiple visits with different campaign parameters.
- A new session is not started with a new campaign parameter.
- A new session is not started at midnight.
- Visiting after the session timeout will start a new session.
Scenarios that are likely to take place:
Scenario 1
- The facebook traffic source would not be counted
- The conversion would be attributed to Google in the User Acquisition view
- The conversion would be attributed to Google in the Traffic Acquisition view
Scenario 2
- The facebook traffic source would not be counted
- The conversion would be attributed to Google in the User Acquisition view
- The conversion would be attributed to Criteo in the Traffic Acquisition view
Scenario 3
All 3 sessions are counted- The conversion would be attributed to Google in the User Acquisition view
- The conversion would be attributed to Criteo in the Traffic Acquisition view
Understanding the Insights
Getting the right insights to understand how channels are performing can vary depending on which reports are being viewed within GA4. This is where a lot of confusion can take place where in GA3 outside of attribution all reports were based on last click which is not the case in GA4.
There are 4 core reports that provide different insights:
- Acquisition > User Acquisition: reports on user first source/medium.
- Acquisition > Traffic Acquisition: reports on session source/medium.
- Advertising > All Channels: reports on the default attribution model that is defined in settings.
- Advertising > Model Comparison: provides a view to compare models from data driven v last click. Last click model is based on Paid + Organic click, it won’t be impacted by the change in how sessions are tracked.
The reports within GA4 provide a view to understand the role of each channel and tactics through a different lens.
Considerations on GA4 sessions data
To be able to report on the correct number of sessions or any other data in GA4 need to consider the following:
- Depending on the volume of traffic and events driven to the website data sampling will be applied to the GA4 data within the interface meaning only a subset of the data will be provided.
- Not possible to get comparable data from GA4 v API v Big Query as its been designed to reduce processing time when dealing with larger data sets.
- Data retention settings will impact the available data within GA4 explorations and API when looking to create a report with data outside the data retention window.
- Google signals data is not exported to Big Query resulting in different outputs between GA4 and Big Query.
Big Query is the most consistent data set available when comparing it to GA4 + API. To ensure there is trust and reliability with the data then Big Query is the best solution available providing huge amount of flexibility.
Case Study
An e-commerce brand was seeing differences in how sessions were being attributed to channels. The big differences were coming from Display, Paid Social and Affiliates for Email, Paid and Organic Search were the big winners. There were significant discrepancies in how channel revenue was being attributed which is a domino impact coming from how sessions are now tracked in GA4 which is at best an estimation on the number of unique sessions ID’s.
(data from Traffic Acquisition, it was the go-to report in GA4 for the e-com brand)
Affiliates is a key channel which had the biggest discrepancy in how sessions and revenue are reported. With GA4 not every session is recognised (especially if it’s multiple visits with different campaign parameters during the same session) unlike GA3. But within the model comparison the performance of Affiliates looked far better than the 60% discrepancy.
Going Forward
The value of a session will be long argued but what Google Analytics (GA4) now provides is multiple views on performance with the various reports and models available. The caveat is understanding the limitations of Google Analytics and working out what is best for the relevant business.
