A Year of Kakobuy Shipping: What Changes Each Season

Three consecutive lunar new year dates — 2026-02-17, 2027-02-06 and 2028-01-26 — move the quietest three weeks of the shipping year across the January to February boundary, which is the reason a shipping calendar cannot be written once and reused. Spring Festival is the only window in the year that stops the pipeline rather than slowing it. Sellers stop shipping and warehouses run reduced hours for roughly three weeks, and the dates move with the lunar calendar, so the pre-holiday month is January in two of those three years and is partly the holiday itself in the third.

In 2026 the holiday begins on 17 February, which leaves the whole of January as a working month and the first half of February as the run-in. In 2027 it begins on 6 February, which shortens the run-in by eleven days. In 2028 it begins on 26 January, so the holiday starts inside January and the run-in sits in the second week of the month. For a buyer the practical difference between those three years is how much slack exists between placing an order and the moment sellers stop dispatching. In 2028 that slack is measured in days rather than weeks.

The January decision is not about price. It is about whether every item in a planned haul can reach the warehouse before the stop. An item ordered in the run-in but dispatched after the last working day does not arrive early; it arrives after the holiday, and it arrives behind everything else that queued in the meantime. A consolidation that depends on one late item therefore does not happen on schedule, and the other items wait with it.

One more feature of the January window matters and is easy to miss. The holiday reduces warehouse hours, so a parcel that is already in storage at the start of the holiday does not move either. If a merge instruction has not been submitted before the stop, it will not be actioned during it. The shipping window calendar instrument holds the five windows against a calendar so that the run-in can be checked before an order is placed rather than after.

The recovery period is a queue-draining period, and queues do not drain linearly. When sellers resume dispatch after roughly three weeks of stopped shipping, the first items to move are not the oldest ones; they are the ones whose sellers restart first, and restart order varies by seller. The same is true at the warehouse, where intake has to process a backlog that accumulated while hours were reduced. A buyer who ordered in the run-in and a buyer who ordered on the first working day after the holiday can therefore end up in the same queue.

What this means in practice is that March and April are poor months for schedule-sensitive orders and reasonable months for everything else. Nothing in the calendar gives these two months a structural disadvantage: the site’s window table contains no entry between the Spring Festival and October Golden Week. The congestion is residual rather than scheduled, which makes it harder to plan around and shorter in duration than the holiday itself.

The one stage a buyer controls directly during a backlog is the review of warehouse photographs. Once the items are in the warehouse, the merge cannot proceed until the buyer has looked at the photographs and confirmed the items, and that step runs on the buyer’s clock rather than the warehouse’s. During a period when every other stage is slower than usual, the review stage is the one that can still be compressed, and compressing it is the only action available that shortens the parcel’s time in storage.

• Expect intake to lag dispatch in March and April, because the warehouse queue drains after the seller queue does.
• Do not treat a slow intake as a lost item. The stage is slower for structural reasons in this period, and the response is to check the warehouse record rather than to reorder.
• Submit review decisions on the same day the photographs appear when the calendar is congested; the alternative is holding a slot in a queue you do not control.

The site’s window table contains no entry between the Spring Festival and October Golden Week, which makes the middle of the year the stretch with the least structural interference of any point in the calendar. Sellers are dispatching on a normal schedule, warehouses are running full hours, and outbound capacity is not competing with a global parcel peak. Nothing in this period is scheduled to be slow.

That absence of a scheduled disruption is the reason this is the stretch in which a haul should be assembled. A buyer who has a fixed deadline has the most room for error here and the least room for error in November. The same order placed in June and in late November faces the same item prices and different pipeline conditions, and the pipeline is the part that cannot be renegotiated after the fact.

Two things do not change with the season. The first is that thresholds and rates are customs parameters rather than calendar ones; a parcel crosses a duty gate in June exactly as it does in December. The second is that each stage of the pipeline still has a friction mechanism attached to it, whether or not a holiday is in progress. A quiet calendar removes added delay; it does not remove the ordinary queue at each handoff.

This is also the period in which parameter corrections are easiest to absorb. The changelog at /vault/changelog/ records status changes, parameter corrections and gaps opened or closed on a weekly basis, so a buyer planning a June or July order has several weeks of log entries available to read before committing. In a congested month the same log is still available, but there is less time to act on it.

Three of the five windows fall inside two months, and they do not share a mechanism. Golden Week and the 11.11 campaign slow seller dispatch. The late-November peak does not slow dispatch at all; it tightens outbound line capacity, so transit stretches even when everything on the seller and warehouse side is running normally. Separating those two mechanisms decides what a buyer can do about each of them.

October Golden Week takes the first week of the month, dispatch resumes the following week, and the mechanism is a one-week national holiday. It is the shortest and most predictable of the five windows: the date is fixed to the calendar by definition, and the recovery is a week rather than a season. An order placed in the second half of September and one placed in the second week of October do not face the same pipeline, and the difference is contained within a fortnight.

The 11.11 campaign sits in early to mid-November and slows seller dispatch for roughly two weeks around it. This is the window with the widest spread of effect, because dispatch speed is a seller behaviour and sellers respond to campaign volume differently. The structural statement that holds across sellers is directional: dispatch slows during the campaign and recovers after it.

The late-November peak is different in kind and is the only window in the year that a buyer cannot schedule around. Capacity is external to the order. Dispatch can be early, review can be fast, and the merge can be submitted the day the items land, and the outbound line still faces the same capacity tightening in the last week of November. The correct response to this window is not to try to move faster inside it but to avoid being inside it when a deadline matters.

• Schedule around Golden Week and 11.11 by ordering before the window or after it; ordering inside a dispatch slowdown gains nothing and costs the same.
• Do not try to schedule around the late-November capacity peak. It affects transit rather than dispatch, and no seller-side action changes it.
• Treat the three windows as separate events when planning a haul. A parcel cannot be inside the 11.11 dispatch slowdown and the late-November capacity peak at the same time, and the responses differ.
• Re-check the shipping window calendar instrument before a November order, because the 11.11 window depends on seller behaviour rather than on a fixed date and is the least predictable of the three dispatch windows.

The year boundary carries two windows at once. The 12.12 campaign sits in mid-December and is shorter than 11.11, with a similar dispatch pattern. The late-November capacity peak can still be unwinding as that campaign begins, because capacity recovers as global parcel volume falls rather than on a fixed date. Both effects are present in the first half of December at the same time, and they stack rather than replace each other.

The second half of December is where the calendar turns and the next constraint appears. The Spring Festival is the only window that stops the pipeline, and it is the next one on the calendar from any point in December. In 2027 it begins on 6 February and in 2028 on 26 January, so the amount of working time between a December order and the holiday stop varies by more than a month depending on the year. A December order that is intended to be consolidated in January should be checked against the lunar new year date for that specific year rather than against the previous year’s experience.

December also raises the storage question in a way that no other month does. Items ordered against a December campaign can arrive at the warehouse close to the January stop, and anything left in storage across the holiday stays there for the duration of it. Every agent warehouse has a free storage window and a charge beyond it, and the terms differ by agent; this site does not publish a verified free-storage figure, so the relevant number is the one in your own account. The structural point holds regardless of the terms: the holiday adds roughly three weeks to any storage period that spans it.

The changelog is the useful tool across this boundary. Weekly entries record which records changed verification status, which parameters were corrected and which data gaps opened or closed, and a buyer carrying a haul across the year boundary is reading exactly the kind of change the log is built to record. The log is a record of changes and not a forecast, and the shipping window calendar is a description of structural timing rather than an observation record. Between them they answer two different questions: what is different from last week, and what is scheduled to be slow this month.