“Session Start (event)” is a metric that Funnel provides when the dimension Event Name and the metric Event Count is included in the connect configuration. “Session Start (event)” aggregates the Event Count of the Event Name “session_start”, which is an event that Google Analytics 4 provides for all properties.
When compared to the normal Session metric, we have found that “Session Start (event)” produces a very similar figure, most often within a 0-5% margin. The totals for “Session Start (event)” in Funnel and “Sessions” you see in the Google Analytics 4 user interface don’t match exactly, but they are very close. Therefore, “Session Start (event)” is useful in most forms of analysis.
What is the challenge with the original Sessions metric?
As you might know, the Google Analytics 4 data model works very differently compared to Universal Analytics. For example, the “Sessions” metric becomes a bit more complicated to work with when exported outside the native Google Analytics 4 interface to a database. This is because Google Analytics 4 is an event-driven analytics platform, and one session typically includes many events.
In most cases, especially if you want to look at granular data, it’s impossible to exactly reproduce the Sessions count you see in Google Analytics 4 in the database. This means that when Event level data together with Sessions is exported from Google Analytics 4 to a database, the aggregated number of Sessions is significantly inflated, sometimes by as much as 5x. This is possible to mitigate by utilising the Event name “session_start” together with the metric Sessions or Event count to create a new custom metric (as explained in the section above).
What can I do if I need even higher precision?
As mentioned, it’s very difficult to get an exact match between Sessions and the Google Analytics 4 user interface and your database. “Session (event based)” will most often get you within a reasonable range of the true Sessions value (0-5% margin). If you want to try to increase the precision, here are two options:
Use the “Sessions” metric instead of “Event count” in the calculation:
You are able to include the Session metric together with the Event name dimension in the connect configuration. Be mindful that this will lead to inflated numbers for the Sessions metric unless properly filtered. However, we have found that the metric totals for Event Name “session_start” together with the Session metric produces results that are slightly closer to the true Session value, typically in a 0-3% margin.
Set up an additional data source with only the Sessions metric and nothing else.
The cause of inflated values in the Session metric is typically related to the inclusion of one or many dimensions in the data source configuration. If you remove all dimensions from the data source selection you are more likely to get results that match or are very close to the true Sessions value in Google Analytics 4.