Page MenuHomePhabricator

[M/L] Deploy Onboarding & "For you" feed revamp as A/B Test
Closed, ResolvedPublic3 Estimated Story Points

Description

Background

We're redesigning the Explore feed: a two-tab experience where 'Community' surfaces Wikipedia's best editorial content in a familiar, scannable format, and For You delivers a full-screen, immersive experience personalized to your interests, reading history, and location.

We need to be able to measure the impact of this improvement on retention and engagement.

Requirements

  • The release with the Explore Feed revamp should only be made availalbe to 50% of iOS app users
    • A: Control (Explore tab, no changes to feed)
      • Existing onboarding
    • B: Variant with redesigned feed
      • Renamed to Home
      • Updated onboarding with Interest selection for new installs and existing app users
      • “For you” feed with new language selector
      • Additional settings for "For you"
      • Existing Explore Feed wrapped into “Community”, With any modules that have been added to "For you" removed
  • Audience: All languages, logged-in and logged-out
  • Analytics must be able to measure retention to the app for Group A and Group B
  • Test should have no end-date. We will create a scaling task to end the test.

Note: follow-up releases will contain

  • Additional “For you” modules (Random, Places, Games), removing from legacy feed as we go
  • Rewritten “Community” tab

Test Notes

Once this work is in, re-test https://phabricator.wikimedia.org/T427577 as well. Complete test notes:

Fresh install latest TestFlight RC. IF:

  • User is assigned control, you will see old legacy app onboarding. Upon completion of onboarding, you should NOT see one-time onboarding in T427577.
  • User is assigned test, you will see new app onboarding. Upon completion of onboarding, you should NOT see one-time onboarding in T427577.

Delete the app, fresh install current App Store App, go through old legacy app onboarding. Then, upgrade to the latest RC via TestFlight. Launch. IF:

  • User is assigned control, you will NOT see one-time onboarding in T427577.
  • User is assigned test, you WILL see one-time onboarding in T427577

Toni Handoff Notes

Based on progress at PR https://github.com/wikimedia/wikipedia-ios/pull/5995

This work has not been done yet.
We won't be adding test kitchen integration.
Just a note that we need to add A/B testing ability into WMFTestKitchen, so this requires some plumbing work before it becomes simple. To build out the original iOS WMFTestKitchen, we asked Claude to reference Android's and port it over to iOS. We can take the same approach to add the experiment pieces to WMFTestKitchen.

Event Timeline

HNordeenWMF triaged this task as Low priority.
HNordeenWMF added a subscriber: Seddon.

@Engineers, and @Seddon please discuss the feasibility of both options here: traditional A/B test vs holdout release

@HNordeenWMF since we won't be using distinct app versions for this AB test we will not be able to identify Uninstalls by App version to differentiate Uninstallers - we can do a pre/post analysis to look at if Uninstalls increased after feature release.

Mazevedo renamed this task from Deploy Explore feed revamp with 50% holdout group, or as A/B Test to [M/L] Deploy Explore feed revamp with 50% holdout group, or as A/B Test.Jul 2 2026, 5:29 PM
Mazevedo subscribed.

M/L - pending deeper review of TestKitchen AB testing capability integration

HNordeenWMF renamed this task from [M/L] Deploy Explore feed revamp with 50% holdout group, or as A/B Test to [M/L] Deploy Explore feed revamp as A/B Test.Jul 2 2026, 8:43 PM

@Mazevedo @Seddon do ya'll need to create a spike for that?

Hi All - I met with Mikhail today and we will not be able to integrate this feature with GrowthBook/TestKitchen for various reasons, will link docs and explanation when I get it from Mikhail. We will just use the experiments."assigned":/experiments."enrolled": columns (Instrumentation) for group assignment values and I will run KR retention analysis manually.

HNordeenWMF raised the priority of this task from Low to Medium.Jul 2 2026, 10:13 PM
HNordeenWMF lowered the priority of this task from Medium to Low.Jul 7 2026, 7:38 PM
HNordeenWMF raised the priority of this task from Low to Medium.Jul 13 2026, 2:46 PM
HNordeenWMF raised the priority of this task from Medium to High.Jul 13 2026, 11:09 PM
HNordeenWMF moved this task from Backlog to P2 (Needs refinement) on the App Experience board.
Mazevedo set the point value for this task to 5.Jul 15 2026, 2:18 PM
Mazevedo changed the point value for this task from 5 to 3.

@GOlson-WMF is there an associated PR with this ticket?

Test plan:

To test updating from old:
Fresh install on MAIN branch (or prior version) OR have prior version installed
Not deleting old version, update to new one (run new branch)
You will be assigned either control or groupB
If you are assigned control:

  • Make sure you see explore as usual, no new onboarding

If you are assigned groupB:

  • Make sure you see the new one time onboarding and are able to go through it
  • Make sure you don’t see it on subsequent installs
  • Make sure you see Community and For You

To test fresh install:
New device/fresh install (fully delete any old versions)
If you're assigned control:

  • Make sure you go through the legacy onboarding
  • Make sure you see explore
  • Make sure you don't see for you/ community

If you're assigned group B

  • Make sure you see the NEW onboarding
  • Make sure you do NOT see the one time onboarding
  • Make sure you see for you/ community / etc
HNordeenWMF renamed this task from [M/L] Deploy Explore feed revamp as A/B Test to [M/L] Deploy Onboarding & "For you" feed revamp as A/B Test.Tue, Aug 18, 5:23 PM

Available in: TestFlight Wikipedia 2026.08.21 (6153)

Tsevener updated the task description. (Show Details)
ABorbaWMF subscribed.

Testing on 6154-beta-2026-08-21
Tested on:
iPhone 16, iOS 26.6
iPhone 11, iOS 26.5
iPad Air, iPadOS 26.6

Requirement✅❓❌Notes
The release with the Explore Feed revamp should only be made availalbe to 50% of iOS app users
A: Control (Explore tab, no changes to feed)
Existing onboardingWorks as expected
B: Variant with redesigned feed
Renamed to HomeWorks as expected
Updated onboarding with Interest selection for new installs and existing app usersWorks as expected
“For you” feed with new language selectorWorks as expected
Additional settings for "For you"Works as expected
Existing Explore Feed wrapped into “Community”, With any modules that have been added to "For you" removedWorks as expected
Audience: All languages, logged-in and logged-outWorks as expected
Analytics must be able to measure retention to the app for Group A and Group BShould be checked by @SNowick_WMF
Test should have no end-date. We will create a scaling task to end the test.Assuming this is good, I can't really check it
Note: follow-up releases will contain
Additional “For you” modules (Random, Places, Games), removing from legacy feed as we goNot present
Rewritten “Community” tabNot completed at this time