Golden Week Capacity: Planning for Concentrated Demand

The timing argument for the Chinese holidays is already settled and we made it in the Chinese shopping calendar: a golden week is when your customer travels, not when they decide, so the work happens months earlier. Take that as read. This post is about the part that comes after you accept it, which is operational and much less discussed. Concentrated demand is not a bigger version of ordinary demand. Everybody wants the same three dates, and the constraint is rarely the one you expect.

For 2026 the shape is worth stating: Mid-Autumn falls on 25–27 September and the National Day holiday runs 1–7 October, with a short gap between them that some travellers will bridge with leave.

The short version: decide four things while it is quiet: which dates you will actually sell, when bookings for them close, what gives when the week gets tight, and whether your published prices hold. Then publish the dates themselves in Chinese, not only the products, because a date marked full answers a question you would otherwise answer by hand. Afterwards, count what you turned away; that number is what tells you whether the constraint was capacity or demand.

What runs out, specifically

"Capacity" is too vague to plan against. Four different things run short, on different timescales, and only one of them can be fixed in the moment.

Departure slots on named dates. Not seats in general. One date fills while the date three days later sits empty, and the traveller asking about the full one has no way to know that.

People. Drivers, guides, and whoever handles a booking that goes wrong. If one person on your team can work comfortably with Chinese-speaking guests, that person's calendar is the real plan, and it is shorter than your vehicle list.

Attention. The enquiries arrive overnight and compete with a week that is already full. A reply window you can hold in March is a different proposition in the first week of October; the mechanics of setting one are in staffing enquiries across time zones.

Things you don't control. Accommodation, permits, timed entry to a site, a boat somebody else operates. For multi-day products these can run out before your own capacity does, and they tend to do it quietly.

Four decisions to make while it's quiet

Which dates you will sell. Go through the window and mark each date open, full, or not running. It reads as administration and it removes more downstream work than anything else on this list, because a date you have decided about is a question you never answer twice.

When bookings close. Past some point a new booking costs more in rearrangement than it earns, and where that point falls depends on how much of your delivery has to be locked in advance. A day tour with spare seats can take a booking the night before; a multi-day itinerary with somebody else's beds in it cannot. Work backwards from what servicing one actually requires: confirming with suppliers, assigning a vehicle or guide, getting documents to the traveller. Where those stop being possible is your cut-off. Publish it, so a late enquiry gets a clean answer instead of an improvised one.

What gives when the week gets tight. Something will: the reply window slips, maintenance waits, somebody's day off moves. Decide which in advance. The reasoning is the same as the high-season failure mode in the staffing post, and the reason to settle it now is that otherwise the choice gets made at eleven at night by whoever is still awake.

Whether prices hold. Charging more for a scarce date is ordinary. Moving a price a traveller has already read is the part that costs you, for the reasons set out in pricing in RMB. If peak pricing is the plan it belongs on the page from the start, next to the deposit and cancellation terms that get tested hardest in exactly this week, which is refunds, deposits and no-shows.

Publish the dates, not only the products

Most operators handle scarcity by saying nothing and answering enquiries one at a time. That holds while the volume is small and stops holding at some point you discover the hard way, and it spends the thing you have least of in a peak, which is your own attention.

Concretely, what belongs in Chinese is a short block on the product screen: each date in the window marked open, full or not running; the date bookings close; the minimum number of people needed to run a departure. Not a booking engine — a list, kept current by you.

The intent is that a traveller who can see a date is full doesn't send the enquiry you were going to decline, and one who can see the next date is open asks about something you can serve. It shifts work from your inbox to a page you write once. Marking a date full is not an admission either; a full date reads as a business that is running.

You keep control of that list, which matters more than it sounds. Availability is yours to set and change, and the point of publishing it is to stop you having to repeat it, not to hand it to anyone else.

The questions that arrive forty times

Concentration makes ambiguity expensive, because the same misunderstanding arrives forty times rather than once. The general case for pre-answering questions on the page is in the storefront anatomy; what is specific to a peak is which questions multiply.

Minimum numbers, and what happens to a date that doesn't reach its minimum. How long you hold a place without a deposit, when every held place has somebody else waiting for it. And what a delayed flight does to a transfer on a day when every vehicle is already committed, which the airport transfer post sets out how to write.

One warning on those. They are the sentences travellers screenshot and quote back at you, so they need writing in Chinese rather than machine-translating. A cancellation rule that comes out slightly wrong in translation is worse than one you never published.

Afterwards, count what you turned away

When the window closes, write down more than the bookings. Count the enquiries, note which dates they asked for, and count how many you declined and why.

The declines are the part nobody records and the part that answers next year's question. Turning away a dozen enquiries for one date tells you which date to open more of. Turning away nobody tells you the constraint was demand rather than capacity, and that is a different problem with a different answer, which we drew out in why translating your website isn't a strategy.

What makes October realistic, or not

Rather than a deadline, the honest version is a list of inputs you can check today. Whether your entity can register and verify the account it needs, and how long that is currently taking. Whether platform review is in front of you, and what tends to send a submission back — that is the review post. Whether the products you would publish have their dates, minimums and terms settled, or whether those decisions are still open.

If those are in hand, the window is workable. If two of them are not, the useful move is to build for the next large window with a proper runway instead of compressing it into the weeks when you can least afford a mistake. Neither answer is a failure; getting to it deliberately is the point.

Where CN1X fits

We build the Chinese-language page or Mini Program that carries the publishable part of the above: the dates with open and full against them, the booking cut-off, the minimums, and the terms that get tested in a peak. You set the availability and change it; we build the surface that shows it. The enquiry reaches the inbox you already use with the date and party size on it.

Pricing, payment, confirmation and the customer stay with you, and we don't promise the demand. Nothing on a page creates travellers who weren't already coming. If the dates you're selling this autumn are settled, tell us what they are and we'll show you what that block looks like in Chinese.