0ec2a40fe0
* Declare winner in update points worker * Remove tests * Test variants for which posts are fetched on home feed * Add field test * Rewrite field tests * Change user to @user * Fix tests * Fix user thing * Check for existence of user before setting experiment * Fix experiment -> @experiment * Change spec * Fix tests * Fix tests * Fix tests * Fix rubocop issues * Remove unneeded offset value * Fix up specs * Final adjustments * Add comments * Update app/services/articles/feeds/large_forem_experimental.rb Co-authored-by: Molly Struve <mollylbs@gmail.com> * Update app/services/articles/feeds/large_forem_experimental.rb Co-authored-by: Molly Struve <mollylbs@gmail.com> * Update app/services/articles/feeds/large_forem_experimental.rb Co-authored-by: Fernando Valverde <fdov88@gmail.com> * Update app/services/articles/feeds/large_forem_experimental.rb Co-authored-by: Fernando Valverde <fdov88@gmail.com> * Better initial user query Co-authored-by: Molly Struve <mollylbs@gmail.com> Co-authored-by: Fernando Valverde <fdov88@gmail.com>
42 lines
1.6 KiB
YAML
42 lines
1.6 KiB
YAML
experiments:
|
|
feed_top_articles_query:
|
|
# This tests how we fetch "globally hot articles" — Testing some variants of
|
|
# more focus on the particpant's tags, some other alternatives which take more into account
|
|
# when the particpant was last on the site and viewed a post.
|
|
# -> Recommended experiment duration: 2 weeks.
|
|
# This experiment has new goals measuring consistently of coming back and clocking a page view
|
|
# and two weeks should be enough to see results.
|
|
# Hypotheses. This will be more impactful than past experiments which were too modest.
|
|
# We'll get more info on bigger levers on experiments, and should be able to pit the winner
|
|
# against some more variation.
|
|
# combination_only_tags_followed_and_top_max_7_days will be interesting to observe I think.
|
|
variants:
|
|
- base
|
|
- base_with_more_articles
|
|
- only_followed_tags
|
|
- top_articles_since_last_pageview_3_days_max
|
|
- top_articles_since_last_pageview_7_days_max
|
|
- combination_only_tags_followed_and_top_max_7_days
|
|
weights: # "base" gets a few more rolls of the dice
|
|
- 20
|
|
- 16
|
|
- 16
|
|
- 16
|
|
- 16
|
|
- 16
|
|
goals:
|
|
- user_creates_comment
|
|
- user_creates_comment_four_days_in_week
|
|
- user_views_article_four_days_in_week
|
|
- user_views_article_four_hours_in_day
|
|
- user_views_article_nine_days_in_two_week
|
|
- user_views_article_twelve_hours_in_five_days
|
|
exclude:
|
|
bots: true
|
|
|
|
# Dashboard performance
|
|
cache: true
|
|
|
|
# For logged-in users we'll explicitely pass participant.
|
|
# If we experiment on non-logged in, we don't want to collect cookies anyway.
|
|
cookies: false
|