Where the bookings came from: 92.2% of 1,650 nights through one platform

Our own three buildings and six rooms sold 1,650 nights over twelve months, and 1,522 of them, 92.2%, arrived through Airbnb (October 2025 to September 2026, measured October 2026, our own reservations). Booking.com brought 126 nights, 7.6%. Direct bookings through our own site brought two nights, 0.1%.

This is not a figure to be proud of. It is the sentence one platform going quiet would take nine out of ten of next month's bookings with it, written as a number.

Which channels do short-term rental bookings actually come from?

Across our six rooms and 1,650 nights in twelve months, 92.2% came through Airbnb, 7.6% through Booking.com and 0.1% direct.

Counted as bookings rather than nights, 438 of 489 came through Airbnb, 50 through Booking.com and one direct. Nights are counted the way the occupancy on our operating data page is counted, one room for one night as one. Direct means our own site's booking engine together with Google Vacation Rentals, which is connected to it, counted as a single channel.

Is channel share measured in nights or in bookings?

Neither figure means much without the other. Over the same twelve months Airbnb was 89.6% of our bookings but 92.2% of our nights, and Booking.com was 10.2% of bookings but only 7.6% of nights.

The gap is stay length. Bookings through Airbnb averaged 3.4 nights, bookings through Booking.com 2.5. A channel that brings shorter stays shows a smaller share of nights than of bookings, and a mix quoted only in nights hides how many extra changeovers came with it.

That gap turns into money. Cleaning is triggered by a changeover, not by a night, so cleaning turns per night sold are 1 divided by 3.4, or 0.29, on the longer channel and 1 divided by 2.5, or 0.40, on the shorter one. The same night sold through the shorter channel carries about 1.4 times the turnovers, and cleaning is the largest line in our published operating costs at 17.0% of revenue (the full cost breakdown).

How long does a second channel take to show up in the numbers?

In our case, seven calendar months: the first night sold through Booking.com fell in March 2026, and by September 2026 that channel was 27.9% of the month's nights.

For the first five months of the window, October 2025 through February 2026, Booking.com sold nothing at all. The climb from there:

That 27.9% is only half good news. Booking.com did climb from 26 nights to 36 that month, but Airbnb fell from 140 nights to 91 in the same month. A share rises when the numerator grows and also when the denominator shrinks, so keep the shares and the night counts side by side.

The slow start has a plain cause: a new listing begins with no track record at all. Photographs and descriptions move across, reviews do not. The work is not the few days of building the listing, it is the months afterwards of keeping the calendar, the rates and the booking conditions from drifting apart across two platforms. Which is why a channel mix that has not moved in the first month of a contract is the normal case, and the judgement belongs to the third month onward (the 42-day median lead time).

Why do Tokyo and Sapporo have different mixes under the same management?

Because the guests are different. Across the same twelve months and the same six rooms, run by the same company the same way, Booking.com was 5.0% of bookings at the three Tokyo rooms and 15.3% at the three Sapporo rooms.

The listing content and the pricing method are the same procedure in both cities. The threefold difference survives anyway, because a channel mix is a result of who travels to that address and what those people already use, not of what the operator prefers. That is why another company's claim of a given direct share is not a forecast for your property: what is worth taking from it is whether the company can measure and produce the figure at all.

How many bookings actually come direct, through the operator's own site?

One booking in twelve months, two nights, 0.1% of nights. Effectively zero.

Our own site carries a booking engine, and guests who book on it pay less than the platform rate. It still produced one booking. The reason is unremarkable: a person looking for somewhere to stay opens a platform, not a search box with our name in it. Direct bookings start to move once a property is searched for by name, and a six-room company needs years rather than months to get there.

We keep the channel open for three reasons. There is no platform commission on it, we can talk to the guest directly before arrival, and it is where a returning guest lands. All three, on today's numbers, sit on the side of what should become true rather than what is true. So when a company sells direct bookings as a strength, ask for the measured share across a named twelve months. If what comes back is not a measurement, it is a plan.

What do extra channels do to commission and to what the owner keeps?

Commission rates move with the channel, the plan and the date the account was opened, so there are no rates printed here. What is useful instead is where to read your own effective rate, which is your own payout statements.

Take one booking per channel and follow it to the end: the total the guest paid, against the amount that reached your account. The gap is not only platform commission. It also holds payment processing, currency conversion, and the lodging tax, whose collection and remittance sit differently depending on the channel (who pays the lodging tax and when).

The second cost of a second channel is not money, it is attention: calendar synchronisation, mismatched cancellation policies and mismatched minimum stays all join the monthly checklist, and a double booking costs far more than the night it lands on (running it yourself against hiring a manager). Adding a channel can also lower the blended commission and give it straight back a few lines down, if the stays it brings are shorter and the changeovers multiply (reading the monthly P and L).

What should I ask a management company about channels?

Four questions. Each has exactly one answer, and a company that cannot answer them is not running those channels.

• ◆The channel mix for a named twelve months, in bookings and in nights. Name the window yourself: a company allowed to choose it will choose a flattering one.
• ◆Whose name the listing accounts are in. The day a contract ends, this is what decides where the reviews and the listings stay (the eight places to read a management contract).
• ◆Nights per booking and cancellation rate, by channel. Both feed straight into cleaning volume and into how often a near date empties.
• ◆How many properties they have actually opened a second channel on, and how long it took. Ours took seven months from the first night to 27.9%.

One question is worth less than it sounds: how many sites are you listed on. A listing count measures work done, not bookings received.

Can this mix be applied to my property?

No. A channel mix is the result of a location, a room size and a guest profile, so another operator's figures are not a forecast for your building.

All six of our rooms operate under a hotel business permit, 旅館業許可, which carries no annual cap on nights (how the permit is obtained). A property staying on the notification route, capped at 180 nights a year, has to decide which demand to fill first, and that changes what each channel is worth to it. A different city splits the way Tokyo at 5.0% and Sapporo at 15.3% split above.

What we can offer an owner is not a future ratio but a measured one every month. Our own figures are published at operating data, and a property we take on gets reporting at the same grain (what belongs in the monthly report). We do not promise an occupancy figure and we do not promise a direct-booking share.