A rug catalog looks small until you do the multiplication. One design in five sizes and four colourways is twenty SKUs — each with its own dimensions, weight, price and imagery, all needing to stay consistent with the other nineteen. A two-hundred-design range is four thousand listings. The catalogs that survive that arithmetic are the ones that made their structural decisions early.
Decide the parent structure first
Which variations live under one parent — size only, or size and colour — is a decision each channel constrains differently, and unwinding it later means rebuilding listings and losing accumulated review history. Decide it per channel, write it down, and never let individual uploads improvise the structure.
Colour names are search data and returns data
'Rust', 'terracotta' and 'burnt orange' are three different search behaviours and one dye lot. The naming decision affects whether buyers find the rug, and whether what arrives matches what the name promised — colour expectation is one of the largest drivers of soft-goods returns. Pick names buyers actually search, keep them consistent across the matrix, and make sure the photography was lit to tell the truth.
Sizes, runners and swatches
Runners and oversizes are merchandising decisions, not just rows in the matrix — they change freight lanes and price architecture. Swatch and sample SKUs are their own small workflow: cheap, margin-irrelevant and disproportionately valuable, because a swatch order is a buying signal worth following up. If samples exist, they belong in the catalog structure from the start.
Twenty SKUs, one story
Copy, materials, construction and care should be written once per design and inherited; dimensions, weight and price must be exact per SKU. Mixing those two categories up — hand-editing shared copy per SKU, or inheriting dimensions that differ per size — is the root of most rug catalog drift.
Watch the drift
Multi-channel rug catalogs drift by default — a price corrected on one channel, a size added on another. A fifty-SKU random sample compared across channels once a month catches it while it is still an errand rather than a project.