Page MenuHomePhabricator

Investigate entry point with unexpected results
Closed, ResolvedPublic

Description

After the initial review of data from entry points for mobile translation (T295757#8256279), there are some interesting cases to investigate further. This ticket proposes to review the instrumentation for the entry points below, check the entry points are shown when intended and the events are fired with the adequate metadata (including user information).

A) Two entry points were identifies with zero events across all wikis:

B) For Tswana (tn) Wikipedia, the "frequent_languages" entry point (which is the most commonly used one in other wikis) has no events (while other entry points do on that wiki).

C) The majority of events (68%) come from users with no previous edits, which is unexpected. This happens across all entry points but especially on the "frequent_languages" one (T298032).

Event Timeline

Pginer-WMF triaged this task as Medium priority.Sep 29 2022, 9:18 AM
Pginer-WMF created this task.
Pginer-WMF renamed this task from Investigate entry point with no instrumentation events to Investigate entry point with unexpected results.Sep 29 2022, 9:46 AM
Pginer-WMF updated the task description. (Show Details)

For case (b) I tested the "mffrequentlanguages" entrypoint (this is the 'campaign' URL parameter for this entrypoint) in production wikis and the event seems to be properly logged with "frequent_languages" event source.

EDIT: As displayed in the screencast, the case (c) also seems to not be caused by some actual issue or bug, as the "frequent_languages" entrypoint is properly displayed for logged-in users and the user information are also properly logged in the "dashboard_open" analytics event.

Additionally, regarding case (a):

  • the issue for the "recent translation" entrypoint is already being handled in T316762, so there is no reason for a duplicate task here.
  • the issue for the "recent edit" entrypoint is indeed still pending, but the original ticket for this entrypoint is still open (T287025), so again I do not think there is a reason for a duplicate task here.

@Pginer-WMF based on my input above, I would suggest that this task can be closed.

@Pginer-WMF based on my input above, I would suggest that this task can be closed.

Thanks for the input, Nik. I added a note on T287025 to make sure instrumentation is verified. All other aspects of the ticket seem clear now. Thanks!