Six months out, hotels have not opened their rooms yet — and it looks exactly like being sold out
We record it when a hotel has nothing available. Reading those records back produced a number that was impossible, which turned out to be the interesting part.
1. The number that did not make sense
Far-out dates looked more sold out than next month
When a hotel returns no room for a date, we record that too — "nothing available" is information, and throwing it away quietly biases every average upward.
Adding it up by how far away the date was gave a result that could not be right: dates six months out looked more sold out than dates two weeks out. Demand does not work backwards.
2. So we asked a different question
Does a given date fill up as it approaches?
Rooms do not un-sell. If a date is genuinely filling, the count of unavailable hotels can only go up. We had been watching the same dates every day since 2026-08-09, so we could simply look.
| Stay date | Day | Days out | 2026-08-09 | 2026-09-06 | Change |
|---|---|---|---|---|---|
| 2026-09-15 | Tue | 9 | 55 | 159 | +189% |
| 2026-09-19 | Sat | 13 | 267 | 432 | +62% |
| 2026-10-17 | Sat | 41 | 180 | 318 | +77% |
| 2026-10-20 | Tue | 44 | 120 | 133 | +11% |
| 2026-11-17 | Tue | 72 | 119 | 123 | +3% |
| 2026-11-21 | Sat | 76 | 241 | 382 | +59% |
| 2026-12-15 | Tue | 100 | 149 | 127 | −15% |
| 2026-12-19 | Sat | 104 | 183 | 214 | +17% |
| 2027-01-16 | Sat | 132 | 274 | 246 | −10% |
| 2027-01-19 | Tue | 135 | 257 | 228 | −11% |
| 2027-02-16 | Tue | 163 | 462 | 292 | −37% |
| 2027-02-20 | Sat | 167 | 468 | 326 | −30% |
Count of hotels returning no availability, out of the same tracked sample in six Japanese cities.
Dates inside three months climbed by a median of +60%. Dates beyond three months moved by −13% — that is, not at all.
3. And one of them went backwards
Which is the part that gives it away
2027-02-16 is 163 days out. Over the same stretch its unavailable count fell from 462 to 292 — −37%.
A sold-out room cannot become available again in bulk. The only thing that explains a drop like that is inventory being opened. Which means those hotels were never sold out. They simply had not started selling that date yet.
4. What this means if you are booking
- Looking six months ahead shows you fewer options, not more. A good share of the market has not opened. The gaps you see are not competition.
- Real scarcity shows up inside three months. That is where the counts actually climb — 2026-09-15 went from 55 to 159 unavailable (+189%) in 30 days.
- We are not saying book late. We cannot say that yet — see below.
5. How this was measured · and what it cannot tell you
- Six Japanese cities, the same tracked hotels, 2 adults, 1 night, asked daily 2026-08-09 to 2026-09-06.
- We count a hotel as unavailable when we ask for that date and it returns no rate.
- Cannot tell you: whether prices are better early or late. Availability and price are different questions; 30 days of history is not enough for the second one.
- Cannot tell you: whether this holds outside Japan. We have the data and have not checked it yet. We would rather say so than imply it.