Background
The Explore feed in the Wikipedia mobile apps was originally designed as an entry point to help readers discover new and relevant content based on the trends of explore feeds more than 5 years ago. However, over time, its utility and appeal have declined due to a combination of shifting user expectations, a static feed, limited visual appeal and other conceptual issues:
- Low engagement: The feed sees minimal repeat use beyond a small, loyal segment of readers with preferences for cards at the top of the feed.
- Outdated interaction patterns: The modular “card” layout feels static compared to modern feeds that emphasize immersive, swipeable, and visually engaging full-screen experiences.
- Lack of clarity and control: Some users are unaware that the feed can be customized or that content is tailored by project and language.
- Limited continuity: Tapping on feed items drops users at the top of articles rather than relevant sections for the user
- “Junk drawer” syndrome: The feed has become a catch-all entry point for unrelated modules (Suggested Edits, Places, etc.) without a coherent information hierarchy or user goal other than promoting the feature
- Untapped potential: The multilingual, multi-project structure could support learning journeys, highlight community impact, and create a deeper sense of belonging within the Wikimedia movement.
Beyond aesthetics, the perceived freshness of the Explore feed directly impacts users’ sense of trust and curiosity. When cards appear outdated, users assume the app itself is neglected or irrelevant. This perception disproportionately affects new and younger readers, who expect constantly updating content streams similar to other modern feeds (e.g., TikTok, Medium, Google and Apple News). This project aims to evolve the Explore feed, the first feature users see when launching the app, into a compelling, modern and movement-aligned experience that encourages curiosity, learning, and daily returns.
Strategic Alignment
The Consumer Strategy calls for reaching new generations of readers and creating memorable, joyful learning experiences that bring them back.
The Explore Feed Refresh directly supports this by:
- Delivering knowledge to more people in engaging, habitual ways. → Makes Wikipedia a place to browse and discover, not just search and leave.
- Supporting “lean-back” exploration and visual learning. → Lets readers enjoy short, useful visits that evolve into deeper exploration.
- Anchoring the “Apps for Deep Engagement” role in the cross-platform journey. → Provides a daily touchpoint for curiosity and community connection within the app.
- Using existing Wikimedia content—text, images, data—in modern formats. → Builds on our strengths, not net-new media types.
Annual Plan
KR 3.1 Engaging new audiences
By the end of Q4, perform at least one experiment per platform (web and apps) that shows a practically significant improvement in logged-out casual reader retention or another indicator metric over control (with casual reader retention defined as 21-day cumulative retention for web, and 14-day cumulative retention for apps)
Overall Hypothesis-
If we redesign the mobile apps Explore Feed we’ll see a 10% increase in Explore Feed engagement over multiple sessions per unique logged-out reader within 14 days of release.
Project Objectives
- Increase engagement frequency: Encourage at least one daily visit to the Explore feed and click through among active app users.
- Modernize the experience: Update interaction patterns to reflect contemporary mobile design standards (e.g., swipeable stories, richer media, seamless previews).
- Clarify personalization: Make customization options intuitive and discoverable.
- Surface impact and participation: Integrate moments that connect readers to the broader Wikimedia movement (e.g., editing impact, community stories, milestones).
- Maintain multi-project, multilingual flexibility: Preserve and strengthen cross-wiki discovery while ensuring coherence.
- Reduce “catch-all” clutter: Create clear governance for what belongs in the Explore feed vs. what should live elsewhere.
Phases
This work will be broken into 4 phases. Subtasks of this epic will include the details of the work of those phases.
Below are the hypotheses for these phases:
- If we investigate how readers currently use and perceive the Explore Feed, we will identify the key usability, relevance and freshness gaps preventing it from becoming a high engagement and retaining feature.
- If we design and test alternative Explore Feed formats, personalization and content types, we will identify the experience patterns most likely to increase user engagement and app retention.
- If we define a coherent governance model and establish the supporting data models and technical underpinnings for an updated Explore Feed, the improvements proposed to the feed will receive positive community and user feedback.
- If we redesign the mobile apps Explore Feed we’ll see a 10% increase in Explore Feed click-throughs over multiple sessions per unique logged-out reader within 20 days of release.
User & Use Cases
Target User
New Reader: Someone who installs or opens the app with intent to read a specific article but has no established habit of returning
Core Use Cases
- “I have a few minutes—what’s interesting right now?”
- “I want to learn something new without committing to a full article.”
- “I want to come back to Wikipedia regularly, but I don’t know where to start.”
Product Principles
Retention before depth
The feed must create reasons to come back regularly, not just read deeply once.
Discovery through remixing, not replacement
Snippets, previews, and remixed content should lead into articles—not replace them.
User-directed, privacy-respecting personalization
Personalization is driven by explicit user choices and transparent signals.
Lightweight engagement that compounds
Small, repeatable interactions (preview, save, follow, dismiss) should gradually shape the experience.
Calm, trustworthy alternative to social feeds
The Explore Feed should feel productive and credible.
Core Functional Requirements
Retaining Factors
- The Explore Feed must:
- Be immediately useful on app open (no empty states)
- Surface a mix of:
- Fresh content (what's new)
- Familiar content (topics users care about )
- Relevant community content (Main Page)
- Engagement encouragement (snippets with scroll, game teasers, previews)
- The feed should:
- Clearly signal "what's new since last visit"
- Encourage short, repeatable sessions (1-5 minutes)
- Optionally allow users to preview, save for later, follow topics and dismiss content. Actions should be reversible and immediately affect feed content.
- Include freshness through language like ("trending" "new this week") and contextual timestamps.
- Highlight updates to followed topics
Success mechanism:
- Users can get value even when they don't have time to read a full article
- Returning feels rewarding, not redundant
Content Remixing
Content Remixing Definition:
- Breaking articles into meaningful, attributable snippets
- Recombining content across articles to surface:
- Related facts
- Thematic groupings
- Contextual pathways
Remixing Requirements
- Snippet-based reading
- Section-level excerpts
- Key facts or images
- Cross-article traversal
- "Related to this" pathways
- Topic-based clusters pulling from multiple articles (articles can be used as seeds)
- Clear Attribution
- Every snippet links back to its source article
- Citations and provenance are never obscured
- Allow users to move:
- Snippet -> Section -> Article
- Snippet -> Related Snippet -> different article
- Encourage curiosity-driven rabbit holes without overwhelming users
- The feed should NOT:
- Present remixed content without clear source context
- Create dead-end experiences that don't lead back to articles for additional reading
Remixing lowers cognitive barriers for new readers and supports exploratory use while reserving Wikipedia's integrity.
Onboarding & Interest-Based Customization
- On first meaningful use, users must be able to:
- Select topics and/or articles they're interested in
- Users should be able to update this setting
- Onboarding should
- Offer popular/trending suggestions (nice to have: localized)
- Allow search by topic and seed articles
- Lightweight and skippable
- Choices should seed feed content and drive remixing and topic clusters
Personalization Model
- Personalization must be:
- Transparent (Because you follow...)
- Based on explicit interests and simple actions (dismiss, save)
- Lightly adapt based on session level behavior
- Personalization must not:
- Rely on black box profiling
- Feel manipulative
Interactive Content
- Games should:
- Be teased in explore feed
- Appear contextually (test your knowledge on this topic)
- Not overpower informational content
Explore Feed Should NOT
- Serve as a social feed with likes, comments and follower counts
- Require account creation
- Summarize content using LLMs
- Feel unstructured
Success Metrics
Key Results
- X% increase in Logged Out readers retained to feature over multiple days
- We see a statistically significant increase in experiment group retained compared to control
- X% increase of unique users engage with Explore feed
- 90 days after release check changes in DAU
Guardrails
- Majority of survey respondents report feeling neutral or satisfied with feature
Curiosities
- Do we see a difference in scroll depth?
- Do we see an increase in article CTR originating from Explore feed?
- Are more users customizing their feed?
- Are we seeing a difference in satisfaction with feature depending on how personalized a user’s feed is?
- What content resonated the most with users?
- Quant: Where did we see engagement?
- Qual: What did users say they’d like to stick around
Release
We are aiming to have an AB test released by May 1 2026