Winner 04 of the finalized 8 (Justin's idea). The path opens with the trip itself: pick a trip type up top like Kayak and Airbnb, then pick your actual cruise from an itinerary list instead of typing ports into a form. Shown with Cruise selected.
Delivering to another ship starts the pattern over for that ship: the Viking Sea now has its own bags-home question. The loop recurses for any chain length with no new UI.
The trip type is the front door (Kayak's vertical tabs, Airbnb's category row as prior art): cruise, golf, ski, camp each get their own path with the right vocabulary and rules; "Anywhere" is the fifth first-class type: plain travel where the product is simply shipping your luggage ahead; it opens the generic path (01's bar). Justin's idea, winner 04 of the finalized 8.
Line, ship, schedule: the port call is the unit of choice (decided with Phil, 2026-08-23). "Departs Sep 2026 · From Barcelona" was an embark framing: the first date shown was when the ship leaves, which is meaningless to someone getting off. Picking the ship and marking calls on its schedule keeps the closed catalog a menu (C2, C3, E021, E011) and makes every date the day the ship is actually in port. Filters are Cruise line, Ship, Traveling (a loose range; an arrives/departs mode select was considered and rejected as re-importing the embark confusion).
Cruise selected by default is defensible: 47.6% of booked trips and 56.6% of value are cruise (E098), and leadership direction is a cruise-specific flow (C67).
The sailing can resolve the amenity: a picked sailing plus a confirmation number joins this path to 02's claim, with the promotion attached before the first price (E024, E099).
Sailing data has real sources (future us): LF already holds partner manifests for line-billed partners (the truest source for Viking-type lines), and the trade offers cruise itinerary APIs: Cruisebound (the cruise OTA that powers Kayak's cruise search), Traveltek (distribution and sailings), Widgety (ship and itinerary content), and cruise modules in the GDSs (e.g. Amadeus). Candidates only; verify coverage and licensing before any build.
Partial cruises are native: each direction is a plain list of the ship's port calls (shown leaving at Livorno, mid-itinerary). Today these bookings are phone-only: back-to-back cruises with different delivery and pickup ports (E009, E020), a two-port itinerary doubling catalog exposure (E021), a bundled guest learning mid-cruise the inbound had to be booked with the outbound (E012). No measured rate yet; a warehouse pull comparing trip ports against sailing embark/disembark ports would give one. Ship-to-ship and onward travel are the next modeling round: repeat the picker block per shipment, same screen, just below.
Two plain questions, one per direction (Phil, 2026-08-23: an interactive mark-the-timeline variant was built and rejected as too complicated; it survives in the kit as .timeline.schedule). "Bags to the ship" and "Bags home from the ship" are each a radio group of port calls (the same pattern as 01's service selection), with "No thanks" as a listed choice and the boarding and leave ports preselected. The other end of either direction can be a street address or another cruise ship, so pickup-from-a-cruise and ship-to-ship back-to-backs are native endpoints, not special flows. Choosing "No thanks" on one side is the one-way case; the baseline's two cruise checkboxes and the direction chips both die.
Each direction is a complete shipment block, exposed one at a time: port call, then the other end, then a priced service. Pricing needs both endpoints (country to country), so the service group can only appear once the port and address are set; sequencing the blocks keeps the price next to the choices that made it.
The pattern recurses per ship (04·3D): delivering to the Viking Sea makes her the current ship, and the same bags-home question runs against her port calls, with the earlier shipments locked above as summary rows. Back-to-backs of any length reuse one screen shape.
Ship-to-ship has one service, not a menu: the arrival is fixed by the destination ship's port call (aboard before the Viking Sea sails), so speed tiers are meaningless and the choice collapses to one confirmed shipment with one price. Service menus only exist when the destination is an address.
The second ship's picker needs no date filter: its port calls are constrained by the first shipment's timing (delivery to the Viking Sea must fall on or after leaving the Star at Livorno, Fri Oct 2), so the schedule shown is already scoped. Only line and ship are asked; the qualifying calls follow.
"Don't see your ship?" keeps the path honest: missing ships and ports force phone calls today (E021); the escape line captures the miss instead of losing the booking.