1884979 - Validate glean.js session ids in the Glean Dictionary against GA

Status: RESOLVED FIXED

(Data Platform and Tools :: Glean: SDK, task, P3)

Description

[Alessio Placitelli [:Dexter]]

In bug 1862955 we added support to counting sessions to glean.js. We recently updated glean.js to the most recent version in the Glean Dictionary. We should measure and understand how the session counts differ (if they do) with GA4.

The GA4 dashboard is available here. To query session counts by day (GA4):

SELECT event_date, count(1) AS nbr_sessions
FROM `moz-fx-data-marketing-prod.analytics_426311622.events_*` 
WHERE _TABLE_SUFFIX BETWEEN '20240301' AND '20240307'
and event_name = 'session_start'
GROUP BY 1
ORDER BY 1 asc

to query page view counts by day (GA4):

SELECT e.value.string_value AS page_location, count(1) AS page_views
FROM `moz-fx-data-marketing-prod.analytics_426311622.events_*` 
join 
unnest(event_params) e 
WHERE _TABLE_SUFFIX BETWEEN '20240301' AND '20240307'
and e.key = 'page_location'
and event_name = 'page_view'
GROUP BY 1 
ORDER BY 2 desc

Comment 1

[Bruno Rosa [:brosa]]

Query: https://sql.telemetry.mozilla.org/queries/99925

I ran this initial query to compare the total number of sessions across the two collection systems. There are more glean sessions because you can block GA data collection. To better understand this, I will log a follow-up bug for the Glean Dictionary where we collect the GA session ID. This will give us additional IDs we can compare and then map to ensure that glean isn't generating too many sessions.

I am going to close this and file the follow-up bug.