Case Study · a concept project
Patreon
Redesigning Patreon's discovery experience so users can find creators — without already knowing who to look for.
Overview
Patreon has tens of thousands of creators — and no way to find them.
Patreon hosts creators across art, music, writing, podcasting, and more — but finding new ones is nearly impossible unless you already know their name. No meaningful browse experience, no recommendation engine, no way to explore by content type.
This concept project redesigned Patreon's discovery from the ground up — adding a home feed, search with filters, and content-first browsing — and validated the design through user testing on Maze.app.
Type
Concept · course project.
Role
Sole designer · full process.
Platform
Web & mobile · responsive.
Testing
Maze.app · real users.
Timeline
1 — 3 months.
The Brief
Design a production-quality feature set — for a real problem, native to Patreon.
A course project with full latitude: research, competitive analysis, wireframes, UI design, responsive layouts, prototyping, and Maze testing — all the way through.
The brief wasn't "design something for Patreon." It was: pick a problem worth solving for a real platform, then go through every step of the design process to a tested, validated solution. The goal was something that felt native to Patreon's existing experience while solving a real, validated problem — not a redesign of the platform but a new section built inside it.
Discovery was the obvious candidate. Every Patreon user I knew complained about it. Every research session confirmed it. The problem was real, well-defined, and shaped by a clear gap in how the existing platform worked.
Problem
Patreon is built for supporting creators you already follow — not finding new ones.
Most Patreon users arrive through external links — a YouTube video, an Instagram post, a tweet. Once inside the app, exploration is essentially impossible.
Problem 01
No meaningful browse experience.
The discover page was minimal at best — a thin grid of creators with no filtering, no categories, and no way to explore by content type or subject matter.
Problem 02
Search required prior knowledge.
Search only returned results if you knew a creator's name or handle. "Digital art" or "true crime podcast" returned little of value — content-first discovery was impossible.
Problem 03
No similar creator recommendations.
Viewing a creator's page offered no pathway to find others like them — no related creators, no suggested content, no hint of further exploration.
Problem 04
Platform growth was capped by external traffic.
Patreon's reliance on off-platform discovery meant creator growth was entirely dependent on social media. The platform itself contributed nothing to helping creators find audiences.
Team & Responsibilities
A concept project — sole designer, real testing.
Concept doesn't mean speculative. The full design process applied — research, competitive analysis, prototyping, and user testing — just without a live engineering team to ship to.
Lead Role
Andrew Fang
Full process — research through testing.
Test Participants
Real users · Maze
Patreon users recruited for user-flow tests.
Inspiration
YouTube · Instagram
Twitch, ArtStation, Twitter — discovery patterns from the platforms users already used.
Reviewer
Course instructor
Critique & review across the design process.
Process
Research built incrementally from four sources
Rather than starting with a full redesign brief, I built understanding incrementally — combining direct research with continuous observation of a live platform in active use.
Source 01
User interviews — real Patreon users.
Interviews revealed a consistent pattern: everyone discovered creators through external platforms — YouTube, Twitter, Instagram, Twitch — then came to Patreon to support them. Nobody had ever discovered a new creator through Patreon itself. Most users did not even know a discover page existed.
Source 02
Competitive analysis — platforms users already use.
Rather than auditing direct Patreon competitors, I studied the platforms users were already on to discover creators: YouTube's recommendation engine, Instagram's Explore, Twitch's categories, ArtStation's content grid. Each offered patterns adaptable to Patreon's creator-focused context.
Phase 1
The key insight from research.
The best discovery experiences lead with content, not creator names. YouTube doesn't ask you to choose a channel — it shows you a video. Instagram doesn't ask you to choose an account — it shows you a post. Patreon's discovery needed to work the same way.
Source 04
The frustration was resignation, not friction.
Users hadn't given up on discovery because it was hard. They'd given up because they'd stopped expecting it to work. The frustration was quiet — invisible in support tickets, invisible in NPS, but loud in interviews once asked directly.
Competitive landscape — five platforms, five capabilities.
Constraints
Designing for both desktop and mobile — the column problem.
Patreon is a web-first platform with a significant mobile user base. Designing discovery meant creating layouts that worked across both — a challenge that went beyond simple responsiveness.
The column count determined the entire feel of the experience. Not just how it looked — how it felt to scroll, how long to find something interesting, whether browsing felt like a chore or a momentum. Resolving the column count became the structural decision the whole project pivoted around.
Source 01
User interviews — real Patreon users.
Interviews revealed a consistent pattern: everyone discovered creators through external platforms — YouTube, Twitter, Instagram, Twitch — then came to Patreon to support them. Nobody had ever discovered a new creator through Patreon itself. Most users did not even know a discover page existed.
Source 01
User interviews — real Patreon users.
Interviews revealed a consistent pattern: everyone discovered creators through external platforms — YouTube, Twitter, Instagram, Twitch — then came to Patreon to support them. Nobody had ever discovered a new creator through Patreon itself. Most users did not even know a discover page existed.
Source 01
User interviews — real Patreon users.
Interviews revealed a consistent pattern: everyone discovered creators through external platforms — YouTube, Twitter, Instagram, Twitch — then came to Patreon to support them. Nobody had ever discovered a new creator through Patreon itself. Most users did not even know a discover page existed.
Source 01
User interviews — real Patreon users.
Interviews revealed a consistent pattern: everyone discovered creators through external platforms — YouTube, Twitter, Instagram, Twitch — then came to Patreon to support them. Nobody had ever discovered a new creator through Patreon itself. Most users did not even know a discover page existed.
The Solution
A content-first discovery system across three connected surfaces.
Not a single screen — a connected set of three surfaces that together gave users multiple entry points into discovery, each designed around a different intent.
Surface
Home discovery feed — content-first browsing.
Cards led with creator content — illustrations, thumbnails, sample posts — rather than names and avatars. Each card previewed what the creator made, with name and category as secondary information. Users could respond to content they found interesting before knowing anything about the creator behind it. Personalised on engaged categories, with a curated section for new and trending creators surfaced at the top.
Surface
Search with filters — find by content, not just name.
Search was redesigned to support content-first queries. "Digital art" or "true crime" returned relevant creators and their work — not just name matches. Filters narrowed by category, tier price range, post frequency, and content type. Patreon became searchable the same way users searched YouTube or Instagram — by interest, not by who they already knew.
Surface
Similar creators — discovery from within a creator page.
A new section on every creator profile surfaced similar creators based on category, content style, and subscriber overlap. Users who landed via external links — the most common entry point — now had a pathway to explore further rather than bouncing back to Instagram or YouTube. Every creator page became a discovery surface, not a dead end.
Surface
Category browsing — explore without intent.
For users without a specific search, a category browse page offered a structured entry point. Categories were visual and content-led — Art, Music, Writing, Podcasts, Gaming — each showing a sample of the work, not just a label. Each led to a filtered version of the discovery feed.
Tradeoffs
The decisions that shaped the discovery experience most.
Four decisions, each with a clear alternative I chose against. The reasoning behind each is what makes them feel earned.
Iteration
What Maze testing confirmed — and what responsive design revealed.
Two iteration loops shaped the final design: real-user testing in Maze, and the column-count exploration that determined the responsive system.
Test outcome
Users found the experience clear, easy to use, and enjoyable.
Maze testing showed users had no significant issues completing discovery tasks — finding a creator by content type, using search filters, and navigating the similar creators section were all completed successfully.
Users reported the experience felt natural and easy to navigate. The content-first approach resonated — several noted it felt more like browsing Instagram or YouTube than using Patreon.
The core design decisions — content-first cards, filter-enabled search, and similar creators — were validated by testing. No major structural changes were required post-testing, confirming the research foundation was solid.
Results
A 0 to 1 product shipped within a year — with thousands of books ready at launch
Kintales launched on iOS and Android with a library of thousands of picture books, each paired with educator-written prompts across four learning subjects. The product went from concept to shipped app within a single year — involving a cross-functional team of designers, developers, educators, and marketing.
Label
1000s
Description
Label
4
Description
Label
0 to 1
Description
Reflections
What the column problem taught me about responsive design.
The biggest challenge on Kintales was not visual — it was deciding whose experience to optimise for when parent and child needs pulled in different directions. A child-first UI risked alienating parents. A parent-first UI risked feeling cold in a moment meant to be warm and playful.
The resolution was to design for parents as the operator and children as the audience. The interface itself was clean and readable for the adult holding the device. The content — prompts, book covers, visual language — was warm and child-appropriate. Once that frame was established, design decisions became much cleaner to make.
Label
Responsive is not just smaller. It's about density.
Layout isn't about fitting content onto a screen — it's about controlling how much a user sees at once and how that affects their behaviour. Too many columns and the experience feels like a catalogue. Too few and it feels like a chore.
The right column count shapes the pace of discovery itself. That's a design decision, not a technical constraint.
Label
Define each breakpoint independently.
Rather than starting with desktop and scaling down, I had to define what the experience should feel like at each breakpoint — and then find the layout that delivered that feeling.
Desktop needed richness and density. Mobile needed speed and momentum. They required different solutions, not the same solution scaled.
Studying platforms users already use.
The competitive research approach on this project — studying platforms users already used rather than direct competitors — was more valuable than a standard competitor audit. Patreon's real competition for discovery attention was not other membership platforms. It was YouTube, Instagram, and Twitch. Understanding what those platforms did well set a much higher bar for the solution.
If I were to repeat this project, I'd run the competitive research even earlier — before the problem framing solidified. Some of the constraints I came to in week six were inevitable once I'd studied YouTube's recommendation engine. The earlier that benchmark lands, the earlier the bar gets raised.