Search at Frontify

Background

Frontify serves 1M+ monthly active users, and search had become critical infrastructure — not just for findability, but for user confidence and trust. Research showed two distinct needs sitting side by side: nearly half of users relied on search when they didn't know where content lived, while the other half used it as a faster shortcut to somewhere they already knew. Search had to support both exploration and efficiency without becoming overwhelming.

The challenge

Before the redesign, search was showing classic signs of a system that had outgrown its design. As content types multiplied — assets, guidelines, text — users lost clarity about what they were even searching. Separate contextual searches fragmented the experience, and exact-match logic with incomplete indexing eroded trust further. More technical power hadn't translated into more confidence: people were disengaging after the first set of results. The real risk wasn't usability — it was that people had stopped trusting search as a reliable decision partner.

My role

I led this end to end as Design Group Lead — problem framing, research direction, interaction and system-level design, and the key UX decisions made across Product, Engineering, and Research. Beyond the screens, I owned how decisions got made, validated, and scaled across teams.

Key decisions

One entry point, not five. Contextual searches forced users to guess where to look before they'd even started. Rather than just improving labeling or fully merging everything into one undifferentiated experience, I introduced a unified, command-style search — one global entry point, accessible via keyboard shortcut from anywhere. The system handles scope complexity; the user just starts typing.

Transparency over "smart but invisible." As indexing expanded, the system could technically surface far more — but people couldn't tell why a result appeared, or why one was missing. We deliberately avoided magical, opaque automation in favor of predictable ranking, clear result labeling, and consistent patterns across content types. The goal was confidence and learnability, not surprise.

Discovery without taking control away. Research showed many users had no idea when editors added or updated content. Search was designed to surface new, relevant content intentionally — without auto-navigating or hijacking intent. It became a collaborative assistant, not an autonomous one.

Accessibility as a system requirement, not a finishing touch. Full keyboard navigation, tested against accessibility standards with external validation, built in from the start alongside Engineering and Research.

Impact

• +10% increase in search engagement, driven by improved trust and clarity
• Higher task completion and deeper engagement with results
• Reduced confusion between content types and scopes
• Search evolved from a background utility into a core, trusted entry point to the platform

System-level outcomes

This work produced more than a redesigned feature: a set of interaction principles and decision frameworks now shared across teams, and a foundation that safely supports the AI-powered and agentic capabilities built since — including the Brand assistant and AI Organiser.