Page MenuHomePhabricator

Weekly updates on editors & readers
Closed, ResolvedPublic

Description

Request for weekly pageview/editing impact report shared within Foundation, in light of changes that may emerge due to COVID-19
Regular analysis of impact (highlevel pageviews & editors) -- weekly analytics report
We theorize editing may go up while people wait around
Who can tell what the future brings?

Requested by @Tnegrin in a Product Leads meeting on 2020-03-13

Event Timeline

kzimmerman moved this task from Triage to Needs Investigation on the Product-Analytics board.

Probably will be assigned to Connie + another person, but we need to evaluate other work that will be pushed back.

I'm going to circle back with Product Leads in the Product Leads meeting Tuesday, Mar 24.

Pageviews are available on a weekly basis in Superset; here's a preliminary dashboard: https://superset.wikimedia.org/superset/dashboard/108/

Editors data is NOT available weekly in the dataset we usually use for reporting on editors (mediawikihistory) - it's only available monthly. We'll have to set aside more time to figure out how to report on editors data on a weekly basis.

Also tagging Analytics so they're aware of this ask.

While data is not nicely denormalized like it is in the monthly dataset, the events database holds every edit/deletion/page creation (each event is its own table) that happens in wikipedia and it is 2 hours behind realtime. Data from those events can be used to estimate editorship on a weekly basis. Now, editor metrics are defined per month (unlike pageviews which are computed hourly), so expect that altering the definition significantly would render data that is not per se comparable.

Hi! I just want to post here to underscore that I would find more frequent edits really valuable. Our team currently waits for monthly updates to Turnilo to look at how our "newcomer tasks" feature has been performing (we look at this view). Looking at it weekly (or daily!) would make a big difference in terms of how quickly we could learn and how much time we spend making separate daily reports in notebooks and spreadsheets.

I posted a similar comment on a similar task: T231938: Get "edits hourly" on a daily basis

@MMiller_WMF We know, really. It is a valuable use case, supporting a denormalized dataset that updates weekly or daily requires a significant project that we are planning on doing in conjunction with core platform and search next year. Infrastructure wise is a significant investment.