Wikipedia Redesign

Redesigning the world's largest encyclopedia for how people actually read

Wikipedia is where most research begins — a first stop for background knowledge, a quick reference mid-conversation, a fallback when other sources fail. Despite its scale and reach, its mobile experience hasn't kept pace with how people consume information: scanning rather than reading, jumping between sections, expecting summaries before committing to depth.

For DSGN 2570 (Design for Usability) at Penn, a complete UX research-to-prototype redesign of Wikipedia's mobile interface was undertaken. The process followed a rigorous human-centered design cycle: structured user interviews, a quantitative survey, synthesis into insight statements and How Might We questions, low-fidelity wireframing, prototype-based user testing with four participants, and a final high-fidelity prototype delivered across five redesigned sections — Home, Article, Search, Language, and an AI Chat feature.

A platform built for encyclopedias, not for people in a hurry

Wikipedia's mobile app functions. But functioning is not the same as working well. Through initial walkthroughs and early conversations, several core friction points emerged — not isolated bugs, but structural mismatches between what the interface offers and how users actually move through information:

• Dense, unstructured text — long articles with minimal visual hierarchy force users to read linearly, even when they only need a specific section. There is no quick summary, no way to preview relevance before committing to a page.
• Broken in-article navigation — users rely on Ctrl+F as a workaround for finding content within a page, a signal that the built-in section navigation is insufficient or invisible to most users.
• Cognitive overload on long articles — technical topics especially sprawl into dozens of sections. Without filtering or summarization, users frequently abandon Wikipedia and search Google instead.
• Buried language and settings — the language toggle is hard to find and poorly labeled, causing confusion for multilingual users who want to switch article language, not app interface language.
• No AI or smart summary layer — other reference tools (Perplexity, Google AI Overview) now surface answers before the user reads. Wikipedia offers nothing equivalent, leaving it feeling increasingly dated for question-oriented searches.

"I usually just Ctrl+F whatever I'm looking for. I don't actually read the article — I just scan for the part I need."

Understanding how three different users actually use Wikipedia

Before designing anything, time was spent understanding how real users interact with Wikipedia — not how they're expected to, but how they actually do. Three structured interviews were conducted across different academic backgrounds: a Wharton Finance student, a College Neuroscience student, and an Engineering Computer Science student.

The discussion guide covered six areas: general usage patterns, navigation behavior, specific pain points, mobile vs. desktop experience, comparison with competing tools, and improvement ideas. A supplementary survey captured quantitative data on usage frequency, session length, and feature awareness.

Key cross-cutting findings from the interview phase:

• All three users primarily use Wikipedia for quick background knowledge — skimming, not reading — and rely on Ctrl+F for in-page navigation because section headers alone aren't enough.
• Long articles with dense terminology cause users to abandon to Google or other tools rather than persisting through the page.
• Users are unaware of or don't use most Wikipedia features (language switching, table of contents, discussion tabs). The features exist but are effectively hidden.
• Mobile usage is notably worse than desktop — smaller targets, more scrolling, and the table of contents is harder to access.

The findings were synthesized into six insight statements and six How Might We questions, each pointing toward a concrete redesign direction.

Mapping the redesigned structure before committing to visuals

Low-fidelity wireframes were produced for nine screens, establishing the core information architecture and navigation model before any visual design began. The lo-fi phase focused on structural decisions: what lives on the home screen, how article navigation is restructured, how the action menu surfaces contextual options, and where language and AI chat fit within the existing tab structure.

Key structural decisions made at this stage:

• Home — redesigned as a search-first surface with trending keywords and today's featured article surfaced as entry points rather than buried below the fold.
• Article navigation — section headers made persistently visible with a scrollable header strip, replacing the buried table of contents. A floating contextual action menu exposes Share, Edit, Discussion, and Page History at the right moment.
• Language — separated from general settings and labeled as "Article Language" to address the confusion between app language and article language.
• AI Chat — introduced as a dedicated tab for users who want answers rather than articles, positioned as a quick-summary layer over the existing knowledge base.

Four users, five tasks, one prototype

Mid-fidelity prototypes were tested with four participants from different Penn schools — Wharton Finance, CAS Biology, SEAS Computer Science, and SEAS Mechanical Engineering — each running through five task-based prompts while thinking aloud. Tasks were designed to test the specific structural decisions made in the lo-fi phase: home page comprehension, article navigation, section search, language switching, and AI Chat usability.

Key observations across all four sessions:

• Home page — all four users understood the redesigned home immediately and found it more approachable than the existing app. The search bar, trending keywords, and today's featured article were interpreted correctly without guidance.
• Article navigation — section headers helped, but users still wanted short previews or summaries before committing to a section. Headers alone weren't enough to decide relevance.
• Search results — users wanted short article snippets in results, not just titles. "I need to know if this is the right article before I click it" was a recurring sentiment.
• Language switching — all four users found the language option but were uncertain whether it would change the article or the app. The label "Language" was ambiguous; "Article Language" resolved this.
• AI Chat — all users found the chat promising but wanted to understand upfront that it produces summaries, not full research answers. The framing needed to be set immediately on entry.

Five sections. One coherent mobile experience.

The high-fidelity prototype delivers a redesigned Wikipedia mobile experience across five core sections, each addressing a specific failure identified in research. The visual language stays close to Wikipedia's established identity — clean typography, minimal chrome — while restructuring the information hierarchy around how people actually scan and navigate.

What habituation hides from usability

The most revealing part of this project was how thoroughly users had adapted to Wikipedia's shortcomings. Ctrl+F for in-page search, a tab open for each related article, Google as the real search layer — these weren't complaints; they were workflows. Users had built systems around the interface's failures so completely that some didn't initially identify them as problems at all.

Surfacing those workarounds as design signals — not just as feedback — was the core methodological challenge. The interviews weren't useful because users said "this is broken." They were useful because of what users showed when asked to complete a task: where they paused, where they reached for something that wasn't there, where they fell back to a behavior the interface hadn't intended.

The AI Chat feature exposed the sharpest tension in the project: adding something new to a platform defined by not being biased or opinionated. Framing it as a summary tool rather than an answer engine — and linking every response back to source articles — was both a usability decision and a philosophical one. Wikipedia's credibility depends on citations and transparency; the chat had to preserve that contract.

The final prototype is a mobile-first redesign, not a feature addition. Most of what changed was not new — it was the same Wikipedia, reorganized around how people actually read.