Hey analytics team! I have a time sensitive question for which I need a decision on Wednesday (turns out this is on the train this week).
tldr: I will lower sampling rate to 1% on English Wikipedia Wednesday PM unless I hear any objections.
Questions to product analytics:
- Will this impact any existing analysis you are doing?
Questions to Analytics:
- What event rate should I be targeting for this schema ?
Background
We recently added an init event to the MobileWebUIClickTracking schema to mirror the init event in the DesktopWebUIClickTracking. This allows us to measure "bounce" rate.
We currently enable the DesktopWebUIClickTracking for 10% of projects in the set of desktop improvements wiki.
We currently enable MobileWebUIClickTracking for 10% on all projects
We can't compare mobile and desktop exactly, but my napkin math [1] suggests that we can expect this to result in at least 1000 events a second (compared to the current 4 events a second) which puts it on par with the VirtualPageViews schema.
I'd like to lower English Wikipedia sampling rate for mobile click tracking from the existing 10% to 1% to be cautious because of this.