Held in one city, the strongest month of the year runs 1.3 to 1.4 times the weakest. In our own numbers the three Tokyo rooms ranged from 72.6% to 95.2% occupancy, a ratio of 1.31, and the three Sapporo rooms from 68.3% to 94.0%, a ratio of 1.38 (the twelve months from August 2025 to July 2026, measured 21 August 2026, method published at operating data).
Average the same twelve months across the whole company and the range narrows to 76.3% to 90.3%, a ratio of 1.18. That difference is not skill. It is Tokyo and Sapporo moving in opposite directions. An owner with one building does not get that cancellation.
When is each city strong, and when is it weak?
Tokyo is strongest in autumn and Sapporo in winter. Grouped into three-month blocks, Tokyo peaks at 90.1% across September to November and bottoms at 81.2% across December to February, while Sapporo peaks at 88.3% across December to February and bottoms at 78.5% across September to November. The two shapes are close to mirror images.
| Three months | All rooms | Tokyo (3) | Sapporo (3) |
|---|---|---|---|
| December to February | 84.4% | 81.2% | 88.3% |
| March to May | 84.4% | 88.8% | 80.0% |
| June to August | 80.8% | 81.8% | 81.1% |
| September to November | 84.8% | 90.1% | 78.5% |
How far apart do the two cities get in a single month?
25.0 points at the widest. In November 2025 the Tokyo rooms ran 93.3% and the Sapporo rooms 68.3%. February 2026 reversed it, Sapporo at 94.0% against Tokyo at 84.5%, 9.5 points the other way. Same company, same manuals, same approach to repricing. The only variable is the city's season.
The troughs also sit in different places. Tokyo's weakest month was August 2025 at 72.6%, then December 2025 at 78.5%. Sapporo's weakest was November 2025 at 68.3%, then April 2026 at 71.1%. Sapporo's low points are not midwinter but the joins between seasons, November and April, once the autumn colours are finished and before the snow arrives, and again once the snow has gone.
Is the company-wide figure steadier because the operating is better?
No. The two cities simply run out of phase. Across the twelve monthly pairs the correlation between Tokyo and Sapporo was minus 0.56 (a correlation coefficient measures how far two series move together: plus one is perfectly together, zero is unrelated, minus one is perfectly opposed).
Put more plainly: in those twelve months there was not a single month in which both cities sat below their own annual average. Whenever one city dipped, the other was up.
That is a portfolio effect, which is what happens when assets that move differently are added together, and not an operating result. An owner borrowing a company-wide average for a single building is planning on a cancellation they do not own. The blended figure describes the company's cash flow, nothing else.
What happens to the cost ratio in a weak month?
Very little. Of the 27.9% of revenue we publish as operating cost, 22.8 points belong to lines that shrink with the nights sold.
| Line | Share of revenue | Behaviour in a weak month |
|---|---|---|
| Cleaning | 17.5% | Falls in proportion to nights |
| Supplies | 4.2% | Falls in proportion to nights |
| Lodging tax | 1.1% | Falls in proportion to nights |
| Electricity | 2.6% | Standing charge stays |
| Gas | 1.0% | Standing charge stays, and winter raises it |
| Maintenance | 1.1% | Not linked to occupancy |
| Insurance | 0.4% | Unchanged |
So where does seasonality actually bite?
Outside that 27.9%. Inside it, the arithmetic moves by a single point.
Sapporo's worst month at 68.3% is, at an unchanged rate, 83.3% of an average month's revenue. The 22.8 points that scale with nights hold their share, while the remaining 5.1 points become 5.1 divided by 0.833, about 6.1% of that month's revenue. Total operating cost goes from 27.9% to roughly 28.9%. One point.
What bites is the spending that does not shrink when revenue does: loan repayments, rent, annual inspections, and in Sapporo the winter preparation, which in our case arrives in full in November, the weakest month of our year (winter in snow country). Note also that the 27.9% excludes rent and management fee by design, and water is absent from the source ledger for this window, so the published figure understates the true total (operating costs, broken out).
How far in advance is a weak month visible?
About six weeks. Across 449 bookings in those twelve months the median gap between booking and arrival was 42 days, so half of next month is already committed today. A weak month is not something that creeps up; it is something already on the screen (booking lead time).
If the same month causes panic every year anyway, the cause is usually not an unwatched calendar but the absence of last year's figure for that month. We keep monthly numbers per city and per room, so from the second year onward the benchmark is that month last year. A first-year owner cannot have that yet, which is why year one is safest treated as the year you record your property's own seasonal shape rather than the year you trade against the market's.
What do we actually do in a trough month?
In this order. Cutting the rate across the board comes last.
- Lower the minimum stay. Vacancy in a weak month appears as one-night and two-night fragments, so removing a two-night minimum alone puts sellable windows back on the calendar.
- Add a long-stay discount for the trough month only. There is a ceiling: past about 8.8% the discount costs more than the changeover it saved (the 3.3-night average).
- Pull inspections, repairs and re-shoots into the trough. Fire equipment checks, furniture replacement, new photography. Closing a room in a strong month is the most expensive way to do any of it.
- Only then touch the rate. Occupancy can be bought by discounting, but a 15% rate cut needs 98.4% occupancy before revenue merely draws level (the occupancy piece).
What should an owner with one building plan against?
That city's worst month, not the annual average. An annual average is a reporting number, not a planning number.
In our own figures, and holding the rate steady, the worst Tokyo month sat at 85.0% of an average Tokyo month's revenue and the worst Sapporo month at 83.3% of an average Sapporo month's. With one building in one city, that trough is yours undiluted.
The sequence is four steps. First, identify the weak season for that city. Second, pencil the weak month's revenue at roughly 85% of an average month. Third, list the outgoings that do not change in that month: repayments, rent, annual inspections, insurance. Fourth, decide how many months of that shortfall you want sitting in the account. Only once the fourth step is decided does seasonality turn from an anxiety into a number (reading the monthly P/L).
What we will and will not promise
We do not tell an owner what percentage their property will run at in a given month. The shape of the year is set by the city, the location and the kind of guest it draws, and an operator moves only part of what sits on top of that. Every figure above is what our own six rooms recorded, not a level being offered for a property we have not seen.
The commitment is to method: monthly numbers per city and per room handed over as they are, set beside the same month last year from the second year onward, and the plan for the trough month built before the trough arrives. Discussing discounts once it has already started is the most expensive way to solve it.