Home & interior: made-to-order and long lead times in feeds and availability data
The second interior headache is temporal: much of the category isn't on a shelf. Sofas made to order in your choice of 40 fabrics, dining tables with a six-week lead time, configured items that cannot be stocked because they don't exist until ordered. Every system downstream of the catalog - feeds, ads, availability displays, customer expectations - was designed around binary in-stock/out-of-stock, and made-to-order breaks the binary. Handled lazily ("everything shows in stock, delivery is a surprise"), it produces the vertical's signature damage: misrepresentation risk on paid channels, and the expectation-mismatch complaints that poison review scores.
Availability as a vocabulary, not a boolean
The modeling move: fulfillment mode and lead time as structured fields - a mode attribute (stocked / made-to-order / preorder / backorder) and a lead-time range (as data: min/max days, not prose), per product or per variant (fabric choices often split modes within one product: standard fabrics stocked, custom fabrics MTO). This vocabulary then renders truthfully everywhere: the product page states the mode and the real window prominently (customers accept long lead times remarkably well when told early - the damage is always the surprise, not the wait), checkout and order confirmations repeat it, and Klaviyo flows (Phase 3) handle the long-wait journey - order confirmation with the window, mid-wait updates, dispatch notice. That journey content is cheap and it's the difference between a six-week wait experienced as craftsmanship versus as silence.
The feed translation: Google's vocabulary for "it takes a while"
Feeds are where the boolean assumption bites hardest, and where the right structured data maps cleanly. Google's availability field isn't binary: preorder (with availability_date) and backorder exist as legal values, and - the underused one - shipping with handling-time attributes (or per-product ships_from / handling windows via shipping settings and, where supported, lead-time attributes) lets a product be honestly "in stock, ships in 4–6 weeks" rather than dishonestly "in stock" or invisibly "out of stock." The mapping rules (Phase 7's transformation layer, earning its keep): fulfillment mode → availability value, lead-time fields → handling/availability dates, and configured-to-order items → either the parent as the feed entity (feed the configurable base, land on the configurator) or excluded from feeds entirely if configuration can't map - a filter rule, decided deliberately per Phase 7's economics rather than left to whatever the default sync does. The compliance stakes are Phase 7's misrepresentation family: delivery-expectation mismatch between ad and reality is exactly the pattern GMC polices; the structured lead-time truth is the immunity.
The demand-capture layer long leads make possible
The reframe that turns this headache into strategy: long lead times mean the order moment and the demand moment separate - and separated demand is capturable data. Preorder and made-to-order flows generate the cleanest demand signal in commerce (paid-for demand, pre-production - the buying team's dream input, extending the Phase 3 back-in-stock logic with money attached); lead-time data structured per fabric/finish reveals which configurations to move into stocked mode (the MTO options customers keep waiting six weeks for are your next stocked SKUs, argued with order data); and honest availability content ("why our sofas take five weeks" - the craftsmanship story) converts the category's structural slowness into brand material while doing the expectation-setting work anyway. The vertical's constraint, wired correctly, becomes its most honest data asset - the recurring shape of this entire phase.