Japan's arrivals fell, but the mix is what changed

The Japan National Tourism Organization estimates 3,098,900 visitor arrivals in August 2026, 9.6% fewer than in August 2025. For January to August the total was 27,626,300, down 2.7%. Those are the two numbers most likely to reach a purchasing meeting, and on their own they read as a soft year.

The country-level detail reads differently. South Korea supplied 850,500 arrivals in August, a record for the month and up 28.7% on August 2025. Taiwan contributed 666,000 (+7.3%), Hong Kong 247,300 (+9.4%) and the United States 197,400. Arrivals from China were 418,000, down 59.0%, and that single market accounts for more than the whole national decline.

A share calculation, shown rather than asserted

Divide the two published cumulative figures and the shift becomes visible: South Korea’s share of January–August arrivals works out to roughly 27% this year, against roughly 22% on the equivalent 2025 figures. That is our division of JNTO’s numbers, not a JNTO statement — worth saying plainly, because a percentage that appears without its arithmetic is a claim, not a measurement.

Why arrivals are not room-nights

A visitor arrival is a border crossing. A disposable towel is consumed in a room, by a guest, on a night, at an occupancy rate that depends on the property, the season and the segment. Between the two sits everything that actually drives replenishment: length of stay, group versus individual booking, the property’s own standard, and whether the guest is one of the roughly 27% arriving from Korea or one of the shrinking share arriving from China.

Treating a 9.6% national decline as a 9.6% decline in towel demand would be a guess dressed as arithmetic. We do not make that guess, and we would push back on a plan that does.

What the mix shift does change

Pack configuration is where this lands first. A hotel buying for a Korean-heavy guest mix and a hotel buying for a Chinese-heavy guest mix can be running the same number of rooms and still be buying different things: different pack counts, different face-towel and bath-towel ratios, and — less obviously — different requirements for what is printed on the pack.

Language on packaging is not a translation problem solved once. It is a per-market decision, and it is cheaper to review it while a quotation is open than after film has been printed.

The second effect is on forecast discipline. A single national growth number, applied across a portfolio, will hide a market that grew 28.7% behind one that fell 59.0%. Where a buyer can share their own occupancy and length-of-stay assumptions, the specification conversation gets shorter; where they cannot, the honest position is a range, not a point.

The question we would ask

Not “how big is the Japan market” but “which guest mix does this property actually serve, and is the current pack built for it”. That question has an answer in the buyer’s own data, and it survives a month like August better than any national total does.