Page MenuHomePhabricator

Export EventLogging data of banner events
Closed, ResolvedPublic3 Estimated Story Points

Description

The Fundraising Operations team needs some event data:

  • All log entries of the "banner-closed" event since Jul 12
  • Log entries of Schema:WMDEBannerSizeIssue during tests 4 to 7; not filtered

Event Timeline

Restricted Application added a subscriber: Aklapper. · View Herald Transcript
kai.nissen set the point value for this task to 3.

Hey @kai.nissen, since this ticket is asking for data which will only be fully collected in the next sprint, I guess this ticket was added to the current sprint by mistake?

Either way and for future reference and in case I get mortally wounded in the meantime, I have amended our infrastructure wiki with a sample query which can be used to query the filtered data, specifically for your second point:

Log entries of Schema:WMDEBannerSizeIssue during test 8 (org-08-181017-ctrl, org-08-181017-var); filtered by user agent (exclude mobile user agents, TV screens)"

See:

https://github.com/wmde/fundraising-infrastructure/wiki/Accessing-EventLogging-data-via-stat1006

@Tim_WMDE I need that data already during this sprint, sorry - except for the data of test 8, of course. Perhaps we split the test 8-part to the next sprint? Data for the other tests could be gathered earliest the day after the start of test 8 (so that test 7 data is complete) which would be Thursday if all holds well.

@Tobias_Schumann_WMDE Ah alright, let's remove the test 8 reference from the requirements then and I will export the rest of the data on Friday.

@Tonina_Zhelyazkova_WMDE Thanks for the export. I have questions regarding the export of the "banner-closed" events:

  • There is no data for destop tests 1-6 in the export and the earliest timestamp is Ocotber 12th. I guess you used incorrect dates in your query. The start date of the export is July 12th (see descpription)
  • The banner-close events for mobile test 2 are very high and the percentage closed banners of banner impressions is so high that it is unplausible. Can it be that 10% instead of 1% of the events are being tracked? (but the export column "event_eventRate" says "0,01")

@Tonina_Zhelyazkova_WMDE Regarding the banner-close events of the mobile test I compared the data to last year and must correct me: the data is not unplausibly high.