Overwolf
Redesigning the CurseForge conversion funnel
Reframed one of gaming's largest modding platforms as a single conversion funnel, then redesigned the path from discovery to download across a surface used by millions of players.
Role
Product Design Lead
Timeline
2025 - 2026
Surfaces
CurseForge website, desktop & mobile web
Scale
130K+ creators · 100M+ downloads · millions of players
Impact at a glance
Mod page, the decision point
Mobile discovery, top of funnel
Browse over Search
Project overview.
The CurseForge website is one of the largest acquisition and conversion surfaces in gaming's modding ecosystem, where millions of players discover mods, decide what to download, and enter the desktop app that anchors the wider CurseForge and Overwolf product.
This work wasn't on a roadmap. The homepage, mod page, and download experience were being treated as separate projects, but users moved through them as one continuous journey, so I reframed the site as a single conversion funnel (Browse → Mod page → Download → App) and used funnel data to show where players were dropping off.
That framing helped align stakeholders and prioritize the work. From there I designed the surfaces as one connected system rather than optimizing them page by page, and carried the work through to launch.
Approach
Navigation & discovery
Making the top of the funnel match how people actually behave.
The Problem
Search was a workaround, not a preference
Browsing made mods hard to evaluate at a glance. Players lacked context within a game's ecosystem, filters were hard to navigate, and thumbnails rarely showed what a mod actually did. So search dominated, while the visual catalog, where engagement actually concentrated, sat below it.
What I decided
Rebuilt discovery, so browsing could carry the decision
A new contextual header improved orientation within a game's content. For browsing itself, I focused on clearer navigation, better filtering, and richer previews so players could assess mods faster. The goal was to make browsing a real decision-making tool, not a path back to search.
Result
Players shifted from searching to browsing
They engaged more with mod content, especially on mobile, where the catalog now led the experience. The visual tile view became one of the most positively received parts of the redesign, giving players a more confident, efficient way to explore before committing.
Outcome
Mobile discovery
Content leads instead of search.
Desktop discovery
Engagement maintained (~44%)
Search/list split clarified intent without costing engagement; window-shopping confirmed dominant.
Install CTA
Clicks −12% · installs −11%
Inconclusive on unintentional installs — flagged for a fast-churn follow-up.
The mod page
Turning the decision point from confusion to clarity.
The Problem
A weak hierarchy and three CTAs competing at the decision point
The mod page is where browsing becomes downloading — the highest-intent step in the funnel — and it worked against the user: a weak hierarchy that forced people to scroll just to orient, and a download path so cluttered you could see three primary buttons competing at once.
What I decided
One download path, built around the three questions players ask
I grouped the page around the three questions a player asks: what is this, can I trust it, is it right for my game. Answering them in order let me pull the key content up and give it real weight. Then I cut the CTAs to one primary path and rewrote the download action so its outcome is clear before the click.
Result
Behavior moved from disoriented to deliberate
In testing, players scrolled to take the page in rather than to find their footing. The drop in clicks and rage-clicks reflected effort moving from understanding the page to evaluating the mod, the right shift on the funnel's highest-intent surface.
Outcome
Mod page
Clearer hierarchy and a single primary download path.
Reflection
The lesson that stuck was about framing. Treating the site as one funnel rather than separate pages changed which problems were worth solving; once the drop-offs were visible at each stage, it was clear where design effort would actually pay off.
It also taught me to match the evidence to the decision. What players said and what they did didn't line up (they reached for search, but engagement lived in browsing), and the most-loved change, the visual tile view, earned its place through demand and reception rather than a single metric. Knowing which signal to trust mattered more than putting a number on everything.
Where I'd push further is precision. With the install CTA I measured total installs when the question was really about accidental ones, so the result was inconclusive; I'd define the exact behavior to move before shipping. And I can show I reduced friction but not yet that I shortened decisions, so decision time is what I'd instrument next.
More projects.
Check out the rest of the featured proojects I've worked on.
