The TestUserSearchSatisfaction schema is currently pulling in about 2k events a day, and with such a high bounce rate to boot that isn't viable for the research we're doing. It's sampling at a 1:1000 rate; dropping that to 1:200 will let us get more data, faster. The stakeholder benefit is we end up being able to explore the user satisfaction hypothesis :).
- Mentioned In
- rMW856c55d0537d: Updated mediawiki/core Project: mediawiki/extensions/WikimediaEvents…
rEWMV4f0bc6277204: Increase search sampling from 1 in 1000 to 1 in 200
rEWMVaf6ef0f258c1: Increase search sampling from 1 in 1000 to 1 in 200
rMEXT345797f7b020: Updated mediawiki/extensions Project: mediawiki/extensions/WikimediaEvents…