When the industry talks about hotel operating costs, the debate has focused almost entirely on unit prices: how much per room to outsource cleaning, the energy charge per kWh, the volumetric water and sewer rate. But cost is unit price × frequency, and frequency is set by the demand structure itself: how many nights each guest stays. At the same 100 rooms and the same occupancy, a market averaging 1.2 nights per stay and one averaging 1.9 nights differ by close to 50% in the number of checkout cleans they generate each year. This article translates that “volume variable” into an operating-cost line item.

HotelBank’s cost-structure series has so far dealt with regional differences in unit prices: the 2.1x gap in water and sewer costs between municipalities, the impact of the 2026 electricity price increase, the per-room burden of fixed asset tax, the surge in outsourced room-cleaning rates, and ADR increases driven by labor-cost pass-through. This article is not about unit prices. It is about turnover — how many nights pass before one checkout clean is triggered. And whereas our earlier analysis of serviced apartment investment economics looked at long-stay assets as such, this article looks at differences in average length of stay across ordinary existing hotels. Previous articles on multi-night stays were written for travelers; this one is written for investors and asset managers.

Average Stay Ranges from 1.171 to 1.857 Nights — a 1.59x Spread Across 47 Prefectures

We calculated average length of stay by prefecture by dividing total guest-nights by the number of guests, using the Japan Tourism Agency’s Overnight Travel Statistics Survey 2025 annual figures (final, published July 6, 2026). The national figure is 1.424 nights. Tokyo is longest at 1.857 nights and Tochigi shortest at 1.171 nights, a 1.59x spread. The median is 1.286 nights: the top three (Tokyo, Okinawa, Kyoto) clearly stand apart, while the remaining 44 prefectures sit in a 1.17–1.49 night band.

The shape of this distribution matters. Average length of stay is not normally distributed; it splits into a small number of long-stay markets and a large number of “essentially one-night” markets. Of the 661.11 million guest-nights nationwide, the markets averaging more than 1.5 nights (Tokyo, Okinawa, Kyoto) account for 174.92 million, just 26.5% of the total. Put the other way, roughly three-quarters of Japan’s lodging market effectively runs on a one-night turnover cycle.

Mapped, the long-stay markets are clearly concentrated in “international gateways” and “islands and resorts.” Tokyo, Kyoto and Osaka accumulate nights as bases for inbound visitors, while Okinawa does so because high travel costs make each trip longer. Conversely, prefectures with hot-spring destinations in northern Kanto, Tokai and Hokuriku are stuck at around 1.2 nights, since the one-night-with-two-meals format dominates.

Guest Mix Holds Up Length of Stay — Japanese 1.320 Nights, Foreign 1.806 Nights

Breaking the national average down by place of residence makes the structure clear. In 2025, Japanese guests averaged 1.320 nights and foreign guests 1.806 nights — a gap of 0.486 nights, or 1.37x. Compared with the previous year (2024), foreign guests edged up from 1.794 to 1.806 nights while Japanese guests edged down from 1.338 to 1.320 nights. The overall figure landed essentially flat, from 1.429 to 1.424 nights.

In other words, the main force holding up the national average is not longer stays within each segment but a rising share of foreign guests, who stay longer. The foreign share of guest-nights rose 2.2pt, from 25.0% in 2024 to 27.2% in 2025, and this mix shift almost exactly offset the shortening of Japanese stays. For investment decisions, note that the cost outlook changes depending on whether average stay is treated as a “permanent characteristic of the market” or as “a variable linked to the inbound share.” How far inbound LOS diverges by prefecture is broken down in Inbound Stay Length Varies 2.0x by Prefecture, down to Tokyo at 2.32 nights and Yamanashi at 1.16.

That said, it would be too simple to read “a higher foreign share means longer stays.” Across the 47 prefectures, the correlation between the foreign share of guest-nights and average LOS is +0.675 — a positive relationship in direction, but with clear outliers. Yamanashi has a foreign share of 27.7%, sixth-highest in Japan, yet its average stay of 1.195 nights is fourth from the bottom, because inbound visitors to the Fuji Five Lakes and Lake Kawaguchi area make near-day-trip one-night loops from Tokyo. Conversely, Kagoshima has a low foreign share of 8.4% but ranks sixth with an average of 1.458 nights, as touring within the prefecture, including its outlying islands, extends stays.

Okinawa is similar. Its foreign share of 27.2% is in line with the national average, but its average stay of 1.792 nights ranks second. Even Japanese guests alone stay 1.729 nights there, more than 0.4 nights above the national Japanese average of 1.320. Length of stay is determined not by the presence of inbound visitors but by travel costs and itinerary design. When using average LOS in investment decisions, treat it not as a proxy for the inbound ratio but as a reflection of each market’s own itinerary structure.

By Property Type: Ryokan 1.196 Nights, City Hotels 1.627 — Turnover Mirrors Product Design

Looking at average LOS by accommodation type reveals an axis separate from regional differences. In the 2025 national figures, ryokan are shortest at 1.196 nights, followed by resort hotels at 1.320, business hotels at 1.443, simple lodgings (hostels and guesthouses) at 1.548, and city hotels at 1.627. The gap between ryokan and city hotels is 0.431 nights, or 1.36x.

Ryokan stays are short because of the product structure itself, which is designed around one night with two meals, including the menu and in-house guest flow. If the same menu is not served on consecutive nights, the trip completes in one night, and in exchange the ryokan can capture a higher rate and more ancillary revenue per night. High turnover is thus a consequence of product design and the flip side of the ryokan asset’s strength. City hotels, by contrast, are built around room-only or breakfast-included stays and have no in-house menu to rotate, so they accommodate multi-night stays more easily. The foreign share of guest-nights also stands out at 47.2% for city hotels, followed by business hotels at 25.6%, resort hotels at 18.3% and ryokan at 12.9%. Which direction the inbound share is moving by property type is tracked with the latest monthly figures in City Hotels: 45.9% of Guest Nights Are Foreign.

The practical implication of this difference is clear. The higher a property type’s turnover, the larger the annual return on efficiency investments per clean (work standardization, cleaning-support systems, better linen transport routes). Ryokan generate the most cleans per room per year of any type, putting them in the position to recoup the same efficiency measures fastest.

Building Up the 100-Room Model — ¥187k per Room a Year Between 1.3 and 2.2 Nights

We now translate this demand structure into operating-cost line items. The base case is a lodging-focused hotel with 100 rooms and 80% occupancy (year-round, model assumption). Annual rooms sold are fixed at 29,200 room-nights, and only average LOS is varied across three levels: 1.3, 1.6 and 2.2 nights. Because occupancy, room count and ADR are all held constant, any resulting difference is purely due to turnover.

The three levels are grounded in actual data. 1.3 nights is almost exactly the national Japanese average of 1.320 and represents a regional-city market centered on business demand. 1.6 nights corresponds to the national city-hotel figure of 1.627 and represents an urban mixed-demand market. 2.2 nights sits between Tokyo city hotels at 2.10 nights and Tokyo’s foreign-guest average of 2.317, representing an inbound base market. All three are observed, real-world levels.

Building up costs on these assumptions, annual housekeeping cost per room (outsourced room cleaning + linen supply) comes to ¥651k at an average stay of 1.3 nights, ¥566k at 1.6 nights and ¥464k at 2.2 nights. The gap between 1.3 and 2.2 nights is ¥187k. For 100 rooms, that is ¥18.7M a year, equivalent to 4.3pt of the model’s ¥438M in room revenue. Per occupied room-night, the figures are ¥2,200, ¥1,900 and ¥1,600.

The cost breakdown shows where the gap arises. At 1.3 nights, 22,462 checkout cleans occur each year, with outsourced cleaning alone costing ¥44.92M. At 2.2 nights, checkout cleans fall to 13,273 and outsourced cleaning drops to ¥26.55M. Stayover nights rise from 6,738 to 15,927, so light-clean costs increase from ¥3.77M to ¥8.92M — but because the light-clean rate is 40% of a checkout clean, the increase comes nowhere near offsetting the decrease. Linen supply moves in the same direction: full linen changes at checkout fall from ¥15.72M to ¥9.29M, while stayover towel changes rise only from ¥0.71M to ¥1.67M.

As a sensitivity analysis, we vary the outsourced cleaning rate by ±20% (¥1,600, ¥2,000, ¥2,400). The higher the rate, the larger the absolute cost advantage of longer stays. At ¥1,600 the gap between 1.3 and 2.2 nights is ¥161k; at ¥2,400 it widens to ¥213k. When outsourced cleaning rates are rising, the relative advantage of owning assets in long-stay markets grows stronger.

Feeding actual prefectural data into the same model allows a more direct comparison. Tokyo, at 1.857 nights, comes to ¥514k per room per year; Tochigi, at 1.171 nights, to ¥702k. The gap of ¥187k is almost the same as in the three-level model comparison. The national median of 1.286 nights sits at ¥656k, and the distribution is heavy at the lower end.

Positioning — Re-sorting Markets into Four Quadrants of Length of Stay × Rate

To put this in a form investors can actually use, we plotted the 47 prefectures with average LOS (2025 final) on the horizontal axis and estimated transacted ADR (June 2026) on the vertical axis. The dividing lines on both axes are the medians: 1.286 nights and an estimated transacted ADR of ¥8,400. Bubble size represents guest-nights.

The upper-right “long stay × high rate” quadrant contains 10 prefectures: Tokyo, Okinawa, Kyoto, Hokkaido, Fukuoka, Ishikawa, Nagano, Niigata, Chiba and Saga. These markets hold a double advantage, favorable on both rate and turnover. The lower-left “short stay × below-median rate” quadrant contains 9 prefectures. Here the turnover burden is relatively heavy — which also means these are the markets where the same measures, such as cleaning efficiency or better linen routes, pay back fastest each year.

What is easily overlooked in practice is the lower right: the 14 prefectures with “long stay × below-median rate,” including Osaka, Aichi, Miyagi, Okayama, Kagoshima, Ibaraki and Aomori. Judged by headline ADR alone they look unremarkable, but because their average stay is above the median, their housekeeping cost is ¥10k–60k lighter per room per year. If this is not reflected in cost assumptions when building the expected NOI at acquisition, the asset ends up being valued on numbers more conservative than reality. Conversely, in the 14 upper-left “short stay × high rate” prefectures, high rates absorb the turnover burden, so maintaining rate is synonymous with maintaining the capacity to absorb costs.

Reflecting It in Investment Decisions — Add One Line to the Due Diligence Assumptions

That this article’s model assumptions are not overly aggressive can be confirmed against listed hotel REITs’ published results. Invincible Investment Corporation’s July 2026 operating results for its 101 domestic hotels were occupancy of 86.3%, ADR of ¥15,100 and RevPAR of ¥13,039 (year on year: occupancy +1.9pt, ADR −1.0%, RevPAR +1.2%). This article’s model uses 80% occupancy (year-round, model assumption) and ADR of ¥15,000, both below those levels.

There are three steps to putting this into practice. First, at the acquisition due diligence stage, add one line to the assumptions table for the average LOS of the market where the target property is located. The Japan Tourism Agency’s prefectural data is free, and quarterly updates are sufficient. Second, hold cost benchmarks for the existing portfolio broken down not as absolute amounts but “per checkout clean” and “per stayover night.” Even at the same annual cost per room, the room for rate negotiation is entirely different when turnover differs. Third, in the design of outsourced cleaning contracts, specify separate rates for checkout cleans and light cleans. In markets where the share of multi-night stays is rising, this two-rate contract is the mechanism that turns the cost advantage into actual numbers.

The same perspective applies to regional banks’ business-viability assessments. When two properties with the same room count and occupancy are compared, a 0.5-night difference in their markets’ average LOS means a housekeeping cost gap of around ¥100k per room per year — about ¥10M a year for 100 rooms. That is not a gap that can be ignored when projecting debt-service capacity.

Conclusion — Put the “Volume Variable” into the Assumptions Table

This article has shown that average length of stay, a demand-structure metric, translates directly into an accounting line item: operating cost. Average LOS across the 47 prefectures ranges from 1.171 to 1.857 nights, a 1.59x spread. Fed into the same model of 100 rooms and 80% occupancy (year-round, model assumption), it produces a gap of ¥187k per room per year in combined outsourced room cleaning and linen supply costs, or 4.3pt of model room revenue. The higher the outsourced cleaning rate, the larger this gap in absolute terms.

The practical meaning comes down to this: long-stay markets have a built-in factor pushing operating costs down. Because it does not show up in headline ADR or RevPAR, it is easily missed at the acquisition stage — which is exactly why it can be a differentiator when properties with the same headline yield are compared. On the other side, short-stay markets are where efficiency investments in cleaning pay back fastest each year, a position of their own in which returns from operational improvement are easier to capture. Since there are levers on both sides, this metric is best seen not as ranking markets but as setting the order in which to pull those levers.

The debate on unit prices has matured. The next thing to put into the assumptions table is turnover.

Related Articles

• Hotel Water & Sewer Costs Vary 2.1x by Municipality: ¥7.85mn NOI Gap
• Cleaning Dispatch Rate Surge & Mid-Size Hotel Pressure: 2026 H2 Polarization
• 2026 Electricity Price Surge: Supply-Side Cost Impact Analysis for Hotel Operations
• Inbound Stay Length Varies 2.0x: Tokyo 2.32 vs Yamanashi 1.16 Nights
• City Hotels: 45.9% of Guest Nights Are Foreign — Ryokan +3.7%
• June 2026 Occupancy 56.2%: Guest-Nights per Room in 47 Prefectures