Chapter 4:
Facet Audit & Search-Refine Prototype
Introduction
Before rolling out my own search-and-refine UI, I audited Zappos’s Women › Flats facet panel to surface taxonomy mismatches and UX pain points. I then built a bare-bones prototype to demonstrate how a fully taxonomy-driven search interface can deliver precise filtering, clear feedback, and no dead ends.
1. Zappos Women’s Flats Facet Audit
Objective
Evaluate how Zappos structures and labels facets for “Women Flats” and identify taxonomy-related issues.
Approach
Captured the collapsed Women › Flats filter panel and two expanded facets (Size and Brand).
Compared ordering, grouping, and scoping logic against the expected taxonomy for a flats category.
Facet panel view (collapsed).
Women Flats “Size” and “Brand” facets.
Key Insights
Scope Mismatch:
Heel Height includes “4 in + Ultra High Heel” yet still returns flats when selected.
Heel height facet misplacement.
Context-scoping glitch
Brand selections replace rather than refine “Women › Flats,” surfacing unrelated apparel.
Brand facet choices replace Flats sub-category.
Over-broad brand list
400+ values overwhelm users despite in-panel search.
Intersection inconsistency
Some filter combinations reset the base category, breaking AND-logic across facets.
Scroll-within-facet UI
Counts and labels hide off-screen, increasing cognitive load.
Missing facets
Key dimensions like Material and Print are absent.
Unremovable base chip
The pre-displayed “Women” chip can’t be cleared.
Product Type misplacement
Shows non-Flats options (e.g. Clothing, Accessories) on a flats page.
Product Type facet misplacement.
Recommendations
Conditional facets
Only surface Heel Height on pages whose class name includes “Heels.”
Scoped Brand list
Limit to footwear brands; add a curated “Popular Brands” subset.
Consolidate dimensions
Merge Product Type into Flats metadata, remove redundant facet.
Related-Subtype facet
Expose hierarchical links (Flats → Loafers, Espadrilles) under “Related Subtypes.”
Value label standardization
Prefix sizes with “US …” and capture EU/UK equivalents as hidden variants.
2. Taxonomy-Driven Search & Refine Prototype
Objective
Showcase a lightweight prototype that applies controlled vocabularies to a clean search-and-refine experience.
Approach
Built a low-fidelity UI that:
Captures keyword queries across all controlled fields.
Shows an “Our Picks” grid on load.
Drives multi-select facets (Color, Material, Print, Sleeve Type) directly from my PoolParty taxonomy.
Displays breadcrumb chips and a “Clear All” button only when filters are active.
Handles zero-hit states with a controlled “No results – try clearing or adjusting your filters…” message plus “Our Picks” best-bets.
Search and refine prototype interaction video. Note that this lo-fi prototype is an 8-item sample to illustrate filtering logic; facet counts (e.g. 12, 20, 15) are drawn from the full catalog metadata.”
Key Insights
1:1 term mapping
UI facet labels (e.g. “Linen,” “Floral,” “Short Sleeves”) match taxonomy
prefLabel.Conditional facet display
Sleeve Type appears only under Dresses or Tops & Bottoms.
Authority-driven best-bets
On zero hits, a taxonomy-managed “Our Picks” list guides users back.
Consistent breadcrumb chips
Each chip (e.g. Color : Beige) uses the controlled “×” removal action.
No dead ends
Users never see an empty grid—recommendations always appear.
4. Next Steps & Enhancements
Pagination/Load more
Add page controls as product counts grow.
Analytics & synonyms
Use search logs to refine synonym sets and surface misspellings.
UX polish
Animate facet toggles, highlight query terms in results, and validate accessibility contrast.
This chapter demonstrates how a rigorous taxonomy audit can inform UI fixes—and how those same controlled vocabularies power a seamless, no-dead-ends search-and-refine prototype.