Marketplace Redesign: Improving Search and Onboarding for a Game Asset Marketplace

Design · 5 min read

Indie developers reported difficulty finding assets: search returned noisy results and category tags were inconsistent because creators used varied naming conventions. The team ran card sorts (N=42) and session recordings to understand mental models. They replaced an open tagging system with a curated taxonomy, introduced faceted filters (engine, file type, polycount, license), and added semantic search powered by embeddings to handle natural-language queries like ‘‘low-poly medieval trees for Unity’’.

The onboarding flow also changed. Previously, new buyers had to create an account before browsing high-resolution previews. The redesign allowed anonymous browsing with high-fidelity previews and instant checkout via ephemeral carts that prompted account creation only at purchase time. Additionally, the marketplace added guided sample packs for common genres and bundled starter kits with clear license explanations.

Post-launch, time-to-first-purchase decreased by 38%, conversion among new indie devs rose 24%, and search satisfaction (measured by successful query-to-download rate) increased substantially. Sellers benefited too: assets that used the curated taxonomy saw higher visibility and sales. The team monitored false-positive search matches and iterated on the embedding model to reduce noise.

This project highlights how taxonomies and progressive onboarding reduce friction in niche marketplaces. Combining structured metadata with semantic search and thoughtful account friction points aligns the product with users’ exploratory behavior and purchasing intent.