Across our own three buildings and six rooms, occupancy ran 83.6% over the twelve months from August 2025 to July 2026: 85.4% for the three Tokyo rooms and 82.0% for the three in Sapporo (measured 21 August 2026, published with the calculation method at operating data).
That 83.6% did not happen in any single one of those months. Month by month the figure runs from 76.3% to 90.3%, a spread of 14.0 points. This article rebuilds the metric from its denominator upward, so that the occupancy claims made by other companies become checkable rather than impressive.
What is occupancy actually dividing by what?
Nights sold divided by nights that could have been sold. The numerator barely moves between operators. The denominator has no industry standard, which is where identical properties end up with different numbers.
Ours is built like this: accepted bookings from our own property management system only, each physical room counted once, and a room counted only from the month it opened. Kameari is listed both as a whole house and as individual rooms, so it is counted on the rooms rather than on the listings. Add listings up instead and the same night is counted twice, which is one of the ways a figure quietly passes 100%.
How far does the 83.6% average swing?
By 14.0 points. The low was 76.3% in August 2025, the high 90.3% in October 2025 (measured 21 August 2026, our three buildings and six rooms over twelve months).
| Month | All rooms | Tokyo (3) | Sapporo (3) |
|---|---|---|---|
| August 2025 | 76.3% | 72.6% | 83.9% |
| September 2025 | 83.3% | 81.7% | 86.7% |
| October 2025 | 90.3% | 95.2% | 80.6% |
| November 2025 | 80.8% | 93.3% | 68.3% |
| December 2025 | 81.9% | 78.5% | 87.1% |
| January 2026 | 81.9% | 80.6% | 83.9% |
| February 2026 | 89.3% | 84.5% | 94.0% |
| March 2026 | 87.6% | 84.9% | 90.3% |
| April 2026 | 80.6% | 90.0% | 71.1% |
| May 2026 | 84.9% | 91.4% | 78.5% |
| June 2026 | 81.7% | 85.6% | 77.8% |
| July 2026 | 84.4% | 87.1% | 81.7% |
Do the two cities move together?
They do not. In November 2025 the Tokyo rooms ran 93.3% and the Sapporo rooms 68.3%: one company, one month, one set of manuals, 25.0 points apart. February 2026 reversed it, Sapporo at 94.0% against Tokyo at 84.5%.
The annual spread is also wider inside each city than across the portfolio: 14.0 points for all six rooms, 22.6 points for Tokyo alone, 25.7 points for Sapporo alone. Combining two cities into one figure makes the average steadier, but that steadiness describes neither city. An owner with one building should be reading the level for their city in that month, not a company average (the Sapporo side of this is in winter in snow country).
Is higher occupancy always better?
No. Occupancy can be bought by lowering the price, so it carries no information without ADR (average daily rate, the mean price of a night actually sold) next to it.
Revenue is occupancy multiplied by rate, so the occupancy at which a discount merely breaks even can be worked out before anybody discounts anything. Starting from our 83.6%:
| Rate cut | Occupancy needed to match revenue | Increase required from 83.6% |
|---|---|---|
| 5% lower | 88.0% | +4.4 points |
| 10% lower | 92.9% | +9.3 points |
| 15% lower | 98.4% | +14.8 points |
| 20% lower | 104.5% | unreachable |
What happens on the cost side when occupancy is bought?
More nights means more changeovers, so the largest cost line grows heavier against revenue. Watching occupancy alone hides this half entirely.
Moving from 83.6% to 95% is 1.136 times the nights. At an average stay of 3.3 nights it is also 1.136 times the changeovers, so cleaning rises from 17.5% of revenue to about 19.9 units against the original revenue. Meanwhile the 15% rate cut has taken revenue down to 96.6, which puts cleaning at roughly 20.6% of revenue. Revenue falls 3.4% and the biggest cost line gains three points of share. That combination is what an owner is looking at when occupancy went up and the money went down (how cleaning runs, reading the monthly P/L).
So what occupancy should an owner aim for?
The target should be nights filled without dropping the rate, not an occupancy percentage. Put a pass mark on occupancy itself and discounting becomes the easiest way to reach it.
The ceiling, though, can be set by the framework rather than by the market. Under 民泊新法, the notification route that caps a property at 180 nights a year, 180 divided by 365 means occupancy tops out at 49.3% even with every permitted night sold. A 旅館業許可 hotel-business permit, which is what all six of our rooms run on, carries no such cap (permit or notification).
Whether a month can still be rescued halfway through is also decided in advance. Across 449 bookings in those twelve months, the median gap between booking and arrival was 42 days. Half of all bookings land more than six weeks out, so next month's occupancy is already half committed. A last-minute discount can only reach the other half, and only on the terms of the table above.
How do you check a company's claim of 95% occupancy?
Ask about the denominator, the period and the scope rather than the number. Without those three, two figures are not comparable. Our own real results can be presented as 90.3% simply by choosing where to cut them.
| Question to ask | Why it changes the number |
|---|---|
| Which twelve months is this? | Cut out one good month or quarter and the same performance reads ten points higher |
| Is this every room you run, or a selection? | One strong room may be telling you about that building rather than that company |
| Does the denominator start at opening or at signing? | Dropping the empty months before launch raises the figure |
| Are owner-use and renovation days inside the denominator? | Excluding them raises the figure, and changes what it means |
| How are rooms listed both whole-house and individually counted? | Adding listings double-counts the same night and can pass 100% |
What we will and will not promise
We do not tell an owner what percentage their property will run at. Occupancy is a composite of location, layout, capacity, season and the rate you set, and an operator owns only part of it. 83.6% is what our own six rooms did, not a level we are offering to reproduce on a property we have not seen.
What we do commit to is method: publishing the calculation and the figure for every room rather than the best one, revising rates daily rather than seasonally, and always reporting occupancy and rate side by side. A report that shows one without the other can be made to look like anything.