Search Experience Rebuild

Search was broken as a product surface: bad states, no business rules, no real tracking loop. Rebuilt it from discovery logic to UX to analytics. Purchases from search doubled.

Context

For a D2C commerce business, search is where customer intent becomes visible, and where the product either earns the next step or loses it. Results quality, empty states, filters, ranking rules, and the tracking behind them all affect whether a session converts.

At Ben & Frank, search had grown without a clear product owner. Decisions about ranking, filters, and UX states had been made in isolation across engineering, design, and marketing. The data layer wasn't set up to support iteration.

Problem

  • No documented business rules for ranking, filtering, or result prioritization. Engineering was guessing at product intent.
  • No-result and low-result states were dead ends. No recovery path, no merchandising logic, no fallback.
  • Tracking reported sessions and clicks but nothing that connected search behavior to purchase outcomes.
  • Four markets with different catalogs, different search patterns, and no shared definition of what good search performance looked like.

Approach

Rebuilt search as a complete product surface. Mapped the customer discovery journey from intent to purchase. Defined business rules for ranking and filtering. Redesigned every UX state: results, empty, partial, suggested. Wired up tracking to close the iteration loop.

Every decision tied to a specific friction point or conversion gap identified in the audit, not a redesign for its own sake.

Decisions

  • Treat no-result and low-result states as product moments, not error screens. Each needed its own merchandising and recovery logic.
  • Align UX patterns across markets while keeping local catalog differences surfaced at the right layer.
  • Build tracking around decisions, not just what customers clicked, but what happened after and why.
  • Partner with marketing and leadership on which discovery improvements should move first based on commercial impact.

Impact

  • 2x purchases from search.
  • A documented set of discovery rules that product and engineering can tune without renegotiating intent every sprint.
  • A tracking layer that supports ongoing iteration, not one-time reporting.