Home & interior: dimensions, materials, and the "will it fit" data problem
Home and interior's defining headache is that its purchase decision is spatial and physical in a way no other vertical matches. The customer isn't asking "do I like it" - they're asking "will it fit through my stairwell, does the seat depth work for my height, will this oak match my floor, does this fabric survive a toddler." Every one of those is a data question, and the vertical's conversion killer is pages that answer none of them. Returns data tells the same story from the other end: wrong-size and wrong-expectation returns dominate the category, and every one was a data gap that shipped.
Dimensions as a system, not a spec line
The core failure is dimensions-as-text: "W180 × D90 × H75 cm" pasted somewhere in the description, format varying by product, packed dimensions absent. The structural version, per Phase 2: dimension metafields with defined semantics - overall W/D/H as numeric fields with units, plus the category-specific set that actually answers "will it fit me": seat height/depth for seating, mattress-fit specs for beds, interior capacity for storage, cable lengths for lighting. Then the two dimension sets almost nobody structures and everybody needs: packed/boxed dimensions and weight (the stairwell-and-doorway question, and non-negotiable for freight quoting and marketplace feeds anyway) and assembly context (assembled vs. flat-pack, box count). What the structure buys, beyond the page itself: dimension filters that work ("sofas under 200cm" - a real filter powered by real numbers, Phase 5's facets in their highest-stakes vertical), feed compliance (marketplaces and Google's home category want structured dimensions), and comparison-shopping presence - dimensional queries ("desk 120cm deep") are precise, high-intent, and only winnable with structured numbers. Consistency doctrine applies with rulers: one unit convention, one measuring convention per category (is sofa depth with or without cushions? - decide once, document, apply), or the filters lie.
Materials and the match question
"Will it match" is a data problem wearing a taste costume. The answerable layer: material and finish as closed-list structured fields (wood species, finish/treatment, fabric composition and - for the quality-signaling that drives this category - the durability data: Martindale rub counts for upholstery, hardness class for flooring adjacent products), color family as its own attribute (distinct from the marketing color name - "Walnut" the name vs. "brown, warm" the family; filters and cross-matching run on the family), and care/durability facts structured (pet-friendly fabric claims, outdoor rating, cleanability). This is also where the vertical's trust content lives or dies: material guides ("choosing between oak and walnut," "what Martindale numbers mean") are Phase 4's content strategy with unusually high purchase-proximity - the customer reading a wood-finish guide is days from a four-figure order.
The "will it fit" content and tooling layer
The structured layer enables the answers customers actually seek, in ascending investment order: dimension diagrams (a drawn diagram with labeled measurements outperforms any spec table - the single highest-ROI content asset in the vertical), room-context photography with scale reference, "measure your space" guides per category (doorway/stairwell checklists for large items - which also measurably cut delivery failures, a cost line finance will notice), and AR/room-planner tooling where catalog scale justifies it (which runs, note, entirely on the dimensional data being right). And the Phase 6 dividend, once more: "will a 220cm sofa fit in a 3.5m room" is exactly the constraint-shaped question customers now ask AI assistants - and the brands whose pages carry extractable numbers are the ones in the answer. In this vertical more than any, the spatial data is the product page.