Background
We are seeing unexpected results from the table of contents A/B test and would like to investigate whether and when there might be issues within our instrumentation (see T309682: Analyze table of contents A/B test for context)
Acceptance criteria
- Review instrumentation for table of contents feature - look in particular for any possibilities for data loss, incorrect bucketing, or other issues
- Review queries from T309682: Analyze table of contents A/B test along with @jwang