Delivery is not a property of a product. The same item, the same seller and the same price will arrive on different days depending on where you are, when you ask, which warehouse holds stock, and what the carrier network is doing that week.
| Input | Why it moves the answer |
|---|---|
| Destination | Distance to the fulfilling location, and whether that lane has same-day service. |
| Order time | Cutoffs. 4:40pm and 5:10pm can be a full day apart. |
| Inventory location | The nearest warehouse may not hold stock; the next one may be 900km away. |
| Season | Peak weeks and public holidays stretch every lane, and estimates get less certain. |
Almost no source exposes it, and it changes the meaning of the promise entirely. “Arrives tomorrow” with nine hours left is an offer. “Arrives tomorrow” with twenty minutes left is a trap you are about to walk a user into.
{ "tier": "express",
"window": { "earliest": "2026-09-09", "latest": "2026-09-09" },
"cutoff": "2026-09-08T21:00:00Z",
"cutoff_in_minutes": 260,
"if_missed": { "earliest": "2026-09-10", "latest": "2026-09-10" } }A single date implies a precision that does not exist. Every estimate is an earliest and a latest with a confidence, derived from observed delivery behaviour on that lane rather than from the seller's own promise — which is optimistic by construction.
Query `nearby` before concluding something cannot arrive in time. Local stock two kilometres away beats every shipping tier, is frequently free, and is invisible to any comparison built on shipping alone.
Around holidays and peak weeks, confidence drops. An agent should widen what it promises rather than repeating a number it no longer trusts — the field exists precisely so that it can.