Plan Search UX

The Challenge

House plans have unusual filter requirements compared to typical e-commerce products. Users care about physical dimensions (width × depth in metres), room counts, number of storeys, and which builder made the plan. The challenge was designing a filter system that felt natural for architects — people who think spatially.

I studied how architects actually browse plans. The key insight: they start with site constraints (block width, then depth), not with the house style. A 12-metre-wide block eliminates most plans immediately. This reversed the typical filter priority — dimensions first, then rooms.

Filter Design Decisions

Range sliders for dimensions, not dropdowns. Block widths vary continuously (8m to 25m+), so discrete buckets would force arbitrary boundaries. Continuous sliders let users set exact constraints matching their site survey.

Bedroom count as toggle chips, not a slider. Unlike dimensions, bedroom count is discrete and low-cardinality (2–6). Chips are faster than a slider for this — one tap vs. drag-and-release.

Builder as a searchable multi-select. With 26+ builders, a flat list would be overwhelming. The searchable dropdown handles both "I want Metricon plans" and "show me everything" use cases.

No "Apply" button. Filters update results in real-time. This gives immediate feedback on how restrictive a filter combination is — if the result count drops to zero, the user sees it instantly and can relax a constraint.

The search interface uses a sidebar filter panel on desktop (always visible) and a bottom sheet on mobile (tap to expand). Results display as cards with the floor plan image prominent — architects need to see the layout at a glance, not read a description.

The filter system handles 1,600+ plans across 13 builders with sub-second response times. The dimension-first approach aligns with how architects actually work, reducing the average filter-to-result time compared to browsing individual builder websites.