Page MenuHomePhabricator

Determine impact of sitemaps on search traffic to Indonesian, Portuguese, Punjabi, Dutch, and Korean Wikipedias
Closed, ResolvedPublic

Description

As a follow-up to T202643, we decided to try sitemaps on a few other wikis (see T206496). This task is about measuring the impact (if any) on search engine-referred traffic on:

  • idwiki
  • itwiki
  • kowiki
  • nlwiki
  • pawiki
  • ptwiki

We have kept the following wikis from receiving the sameAs A/B test (T208755) to use as "controls" when doing the analysis using the CausalImpact package:

  • Ukrainian (0.765 w/ Indonesian, 0.699 w/ Portuguese)
  • Bhojpuri: bhwiki (0.77 w/ Punjabi)
  • Cherokee: chrwiki (0.75 w/ Punjabi)
  • Kazakh: kkwiki (0.7 w/ Portuguese)
  • Catalan: cawiki (0.64 w/ Dutch)
  • French: frwiki (0.63 w/ Dutch)
  • Yoruba: yowiki (0.55 w/ Korean)
  • Kalmyk: xalwiki (0.51 w/ Korean)

(naively chosen via Pearson correlation of daily pageviews from September & October)

Event Timeline

mpopov triaged this task as Medium priority.Nov 16 2018, 5:56 PM
mpopov created this task.
Restricted Application added subscribers: Base, revi. · View Herald TranscriptNov 16 2018, 5:56 PM
mpopov moved this task from Triage to Next Up on the Product-Analytics board.Jan 10 2019, 8:08 PM
mpopov moved this task from Next Up to Doing on the Product-Analytics board.EditedFeb 7 2019, 4:29 PM

Just like with sameAs (T211191#4885005), there is no visible change in traffic due to the intervention:

@ovasileva There's a slight bump after deployment but there's a bump in the previous years too, so it's hard to tell with a naked eye whether that's attributable to the deployment or not. The impact looks flat at worst and slightly positive at best. As such, I don't think a rigorous, statistical analysis is needed here because there's no reason to suspect there was a negative effect.