Holiday Bookings Are Arriving Early. Should a Hotel Cut Service First, or Segment It First?

China's 2026 Mid-Autumn Festival and National Day are separated by only three working days.

Under the General Office of the State Council notice issued on 4 November 2025 (Guo Ban Fa Ming Dian [2025] No. 7), Mid-Autumn 2026 runs from Friday 25 September to Sunday 27 September, with no shift adjustments, and National Day runs from Thursday 1 October to Wednesday 7 October, with two surrounding weekend make-up working days: Sunday 20 September and Saturday 10 October. If a property owner and general manager add three days of paid leave across the gap (28, 29 and 30 September), a continuous 13-day window becomes possible.

Bookings are already inside the system. According to the Umetrip (航旅纵横) civil aviation official direct-sales platform, as cited by China Civil Aviation News / 《中国民航报》 on 13 August 2026 (reporter Chen Hongying), by 11 August the domestic route bookings for the holiday window (25 September to 7 October) had passed 1.71 million, up roughly 7% year on year, and inbound-outbound bookings had passed 1.28 million, up about 11% year on year. The same report also references Tongcheng Travel data showing that in the week of 31 July to 4 August the search heat for holiday-window outbound flights doubled week-on-week, and outbound tour package bookings rose roughly 60% week-on-week.

The hotel-side industry signal is mixed. The China Hotel Association's Lodging Consumption Index Report (July 2026), released in August 2026, shows the composite China Hotel Industry Consumption Index (HCI) at 97.3 in July 2026, up 9% month on month and 1.6% year on year; the average room rate index fell 5.9% year on year, the occupancy index fell 1.3% year on year, the food and beverage revenue index fell 17.9% year on year, and the online booking index rose 10.1% year on year. The report frames the data carefully: peak-season recovery, price pressure and food-and-beverage pressure can occur together; an annual plan cannot be written around a single aggregate growth rate. The index uses May 2025 as the base of 100.

Early demand is not early experience. The real structure of bookings, the real needs of guests, and the real carrying capacity of the property are not yet fully visible in the system. They will only become substantially clearer by mid-to-late September.

The first meeting a general manager gets wrong

It is not the price meeting. It is the "do we have enough hands" meeting. Both questions are urgent, but the order matters. If headcount is discussed before service is segmented, the answer becomes a blunt plus-or-minus on total labour. If service is segmented first, the conversation becomes which moves are necessary, which can be substituted, and which can be flattened.

The answer first

Do not cut service as a whole. Segment service into three layers before any decision.

The first layer is non-negotiable promises. Sold breakfast, transfers, child or elder assistance, key group-arrival moments, confirmed room features and late check-out are commitments the guest has already paid for or been explicitly told. They are not service extras. They are the credit line of the property, and each must have a named owner and a documented fallback.

The second layer is experience that can be booked or smoothed. Breakfast, concierge support, transfers, selected activities and local experiences are often not impossible, only impossible for everyone at the same time. Pre-arrival information, time-slot booking, area routing and clear on-site guidance can move part of the peak away from the front desk and the restaurant without removing the experience.

The third layer is convenience that can be substituted. Digital guidance, partner services, time adjustments and standardised self-service tools can take on a share of the work. They cannot replace safety, hygiene, accurate delivery or emergency response.

Mixing the three layers in one meeting creates two recurring errors: cutting committed service to save cost, or carrying every non-essential service on the peak shift to avoid complaints.

Two things worth doing before the holiday

First, align sold promises with actual workload rather than aligning headcount with occupancy.

Occupancy tells a property how many rooms it sold. It does not tell the team when work will arrive or how complex it will be. Two properties can both run at 85% occupancy. One may have stable leisure long-stays; the other may have one-night business guests, families, groups and sold benefits arriving in the same two hours. Staffing them by the same percentage is not precision. It is risk pushed to the front line.

Before the peak period, four questions belong on the operating stand-up agenda more than "do we need another body":

1. Which arrival dates, arrival windows and departure windows in the next 14 days are actually congested;
2. Which reservations carry non-negotiable promises or special-assistance needs;
3. Which services can be time-slotted, redirected or explained before arrival;
4. If a step falls through, who can reallocate people, approve outside support, or authorise recovery.

Second, run a 20-reservation promise walk-through for the next 14 days.

No new system is required. Only a six-column worksheet.

Column one records the sold promise, what the guest has paid for or been told. Column two records the service touchpoints, where front office, housekeeping, food and beverage, concierge and engineering each engage. Column three records the time window, when check-in, breakfast, meetings and check-out peak. Column four records role accountability, who must deliver and who can support. Column five records the fallback plan, what happens when hands are short, the room is not ready, or the restaurant is full. Column six records the recovery boundary, who can authorise compensation or substitution and under what triggers.

The value of the worksheet is not to personalise every reservation. It is to surface the gap between what commercial has sold and what operations has prepared to deliver. Once the gaps are visible, fix handovers and priorities before touching headcount.

Who moves first, and by which yardstick

Owners set the guest-experience floor. Which services cannot be cancelled no matter how busy, and which exceptions can draw on a pre-approved budget, must be stated by the owner and general manager in advance. Otherwise department heads are forced to make trade-offs in front of the guest.

General managers translate the priority into shift design. Not vague "we care about service", but specific answers: who supports the breakfast peak, who coordinates late room releases, who greets groups ahead of arrival, who owns the escalation path when a complaint is filed.

Department heads adjust the next day by the previous day's numbers. Promise completion rate, check-in wait time, room-release timeliness, breakfast queue length, first-touch 24-hour complaint count and overtime hours. Watching only the labour cost ratio teaches the team to compress cost without stabilising service.

Pre-arrival smoothing should also be live 2–3 weeks before the peak. A confirmation 24 hours before arrival that states breakfast windows and check-in flow suggestions; a three-queue front desk during peak (standard, member, group); a concierge desk that separates split routing, luggage, enquiries and ticketing; a housekeeping schedule driven by arrival and departure windows rather than by floor. None of these are large projects, but each needs lead time.

Replace the meeting debate with a 14-day minimum-cost test

Pick one unstable promise as the test subject. Three selection rules: one promise only, one with comparable observation days, one that can be reversed after the test.

Days 1 to 7 hold staffing and current action steady under one common rubric and record the baseline. Days 8 to 14 change one low-cost variable at a time. Examples: the pre-arrival confirmation template, the front desk routing script, the housekeeping cleaning order, the restaurant recommendation language, the group arrival flow. Do not change decor, pricing, large spend or vendor contracts. Hold the weekday mix constant, define the eligible denominator before comparing, then look at promise completion rate, room-release timeliness, check-in wait time, breakfast queue length, first-touch 24-hour complaint count and overtime hours.

Write the conditions to continue, adjust and stop before the test starts. For instance, if promise completion does not improve across three comparable observation days, stop layering changes and return to the 20-reservation walk-through for re-diagnosis. If complaints or safety risk rise, revert immediately. If a sold benefit is repeatedly missed, escalate it into the first layer under direct general-manager ownership.

The 14-day test exists to filter one break-point for the next round of repair and one inefficient action to pause or remove. It does not by itself prove long-run causation, and it does not authorise a dramatic headcount cut or product redesign on its own.

Boundary and counter-cases

This approach fits established properties with usable reservation and handover records. New openings, major-event properties and destinations exposed to weather or safety risk should shorten the review cycle to intraday or hourly and raise general-manager involvement. Properties at full occupancy with a simple customer mix do not need a complex version of this framework. The real task is to identify the actual commitments and the actual peaks.

Two moves to make tomorrow

Before tomorrow morning's stand-up, open the next 14 days of reservations and pick 20 representative ones. Complete the six-column walk-through for each. In the afternoon, hand the result to the general manager and re-sort each line into non-negotiable, smoothable, or substitutable.

Only after a property knows what it must protect can it decide where to be lighter, faster and leaner.

MarvelBros C&T