Line up guest review scores for accommodations across Japan, and most analyses collapse into a single yardstick: which prefecture scores highest. But the guest experience is not one-dimensional. A 4.2 earned on the strength of the bath and a 4.2 earned on the strength of location are entirely different businesses. For this article we pulled the guest review baseline for all 47 prefectures and decomposed each prefecture’s ratings into “how much praise it receives” (level) and “where the praise lands” (relative profile). What emerged from 16,007,660 reviews (N=47 prefectures) is that the 47 prefectures fall into four clearly distinguishable types — and that “where the praise lands” has almost nothing to do with room rates.

Food is the most praised nationally; bathroom has the most headroom

First, the national baseline that underpins everything. Taking a simple average of each of the 47 prefectures’ category-level mean scores, the highest is Food at 4.314, followed by Staff at 4.268 and Breakfast at 4.213. At the low end are Bathroom at 3.958, Facilities at 3.972 and Rooms at 4.049. Japanese accommodations share a common national structure: what is served and the people serving it are rated highly, while the building itself rates below them.

Source: HotelBank Editorial Team research (aggregation of guest reviews, N=47 prefectures, 16,007,660 reviews, most recent 24 months)

What matters here is less the average itself than the dispersion. Bathroom’s standard deviation across prefectures is 0.144, by far the largest of any category and roughly 1.5 times that of second-placed Food (0.097). Between the highest, Gunma at 4.233, and the lowest, Fukuoka at 3.699, lies a gap of 0.534 points — wider than the prefectural spread of any other category. In other words, Bathroom has the lowest national average, yet is simultaneously the axis where a prefecture’s character shows most clearly. In prefectures with hot spring destinations, large communal baths and open-air baths draw high ratings; in city-centric prefectures, the in-room unit bath becomes the focus of evaluation. The same word “bathroom” points to entirely different facilities depending on the prefecture.

This analysis covers ten categories, but among them Amenities falls below 30 covered properties in 18 of the 47 prefectures, leaving a thin base. It is therefore excluded from the type analysis below and shown only as a national average for reference. Service and Staff also showed a tendency for ratings to skew toward one axis or the other depending on the prefecture (a correlation was observed between the ratio of covered property counts for the two and the difference in their relative scores), so for the type analysis the two were merged into a single axis, “Hospitality”, weighted by property count.

Source: HotelBank Editorial Team research (aggregation of guest reviews)

Measuring “where the praise lands” — the relative profile, stripped of level

Rank prefectures by raw scores and the top of the list fills with tourism prefectures: Kyoto (eight-axis average 4.298), Oita (4.269), Nara (4.253). That is a statement about level — that a prefecture’s properties are highly rated overall — and useful in its own right, but management decisions require one further decomposition. Every prefecture, without exception, has axes that are relatively strong within it and axes with relative headroom.

So from each prefecture’s category scores we subtracted both that prefecture’s overall level and that category’s national level. What remains is the upside or downside specific to that prefecture on that axis alone. The vector of these residuals across the eight axes is that prefecture’s relative profile.

Running principal component analysis on this eight-axis data, the first principal component alone explained 52.3% of total variance. Its content is unambiguous: on the positive side sit Food (+0.41), Bathroom (+0.36) and Breakfast (+0.35); on the negative side, Rooms (−0.44), Facilities (−0.39) and Location (−0.39). In short, the largest single divide in Japan’s accommodation ratings is whether you are praised for bath and food, or for the building and the location. This can be read as the difference between ryokan/hot-spring-style stays and city hotel stays showing up directly as a difference in rating axes.

The 47 prefectures split into four types

Clustering the 47 prefectures on their eight-axis relative profiles produced four highly interpretable types. Here is the overall picture.

Source: HotelBank Editorial Team research (aggregation of guest reviews)

Source: HotelBank Editorial Team research (review aggregation) / MetroEngines Research (estimated settled ADR)

The types separate cleanly on the map

Plotting the four types on a map makes their geographic clustering visible. Bath type concentrates in the hot spring belts running from Tohoku through northern Kanto, Shinetsu and Chubu, plus central Kyushu. Hardware (rooms & facilities) type is distributed across Hokkaido and Okinawa and the urban areas of the Seto Inland Sea and Hokuriku. Food type is common from the Kii Peninsula through San’in, Shikoku and southern Kyushu — the country’s well-known ingredient-producing regions.

Source: HotelBank Editorial Team research (review aggregation) / MetroEngines Research (estimated settled ADR)

Bath type — 17 prefectures, the broadest type

The largest type is Bath type, with 17 prefectures. Oita, Shizuoka, Gunma, Nagano, Gifu, Tochigi, Yamagata, Fukushima, Iwate, Akita and other prefectures with well-known hot spring destinations fall here almost without exception. In the relative profile, Bathroom stands out at +0.99, followed by Breakfast (+0.40) and Food (+0.30). Location (−0.94) and Rooms (−0.74), meanwhile, carry relative headroom. Hot spring properties are often located away from station fronts and downtown districts, and ratings gather around shared experiences — the large communal bath and the meals — rather than the guest room on its own. On how properties offset that locational handicap, Hotel Shuttle Reviews: Airport Shuttle vs Onsen Station Pickup Types digs into the moves available on the hot spring side.

Worth noting is that this type’s median estimated settled ADR of ¥10,700 places it second only to Metropolitan type. Its eight-axis average score of 4.174 is also the highest of the four types. Prefectures where ratings gather around bath and food sit at the top on both rate and satisfaction, despite lacking the locational advantage of the big cities. This suggests that in domains where the experience itself is the product, a locational handicap has little bearing on price.

Metropolitan type — 3 prefectures, with ratings centered on building and cleanliness

Kyoto, Tokyo and Osaka alone form an independent type. Their relative profile is strong across the board on Cleanliness (+2.27), Rooms (+2.01), Location (+1.85) and Facilities (+1.76), while Food (−2.62), Breakfast (−1.73) and Bathroom (−1.60) retain relative headroom. The amplitude is an order of magnitude larger than in the other 44 prefectures, which shows just how far these three depart from the national average picture.

Median estimated settled ADR is ¥14,700, the highest of the four types. Within it, though, the spread is wide: Kyoto ¥16,700, Tokyo ¥14,700, Osaka ¥11,000. Room-only and breakfast-only stays dominate, and stays where dining is not completed on-property are the norm — a structure that likely sits behind the relative weakness of the food axis. Turned around, properties in these three prefectures that can build up ratings for on-property dining can occupy a position that is scarce even by national standards.

Hardware type (15) and Food type (12) — two characters splitting the mid price band

The remaining 27 prefectures divide into two types whose median estimated settled ADR sits neck and neck in the ¥8,000s. Hardware (rooms & facilities) type has 15 prefectures, relatively strong on Rooms (+0.81), Facilities (+0.66) and Location (+0.53). Kanagawa, Fukuoka, Okinawa, Ishikawa, Hokkaido, Aichi and Hiroshima line up here — prefectures anchored by regional core cities and resort hubs. Many are areas where recent openings and renovations have been concentrated, and the newness of buildings and rooms appears to be feeding through into ratings.

Food type has 12 prefectures, relatively strong on Food (+0.90), Breakfast (+0.67) and Hospitality (+0.22). Nara, Mie, Fukui, Wakayama, Tottori, Kochi, Miyazaki, Kagoshima and Tokushima — prefectures known as ingredient-producing regions — are at its center. The highest food score in the country is Nara’s 4.481, and the highest breakfast score is likewise Nara’s 4.372, both national number ones. This type carries relative headroom on Cleanliness (−0.86) and Rooms (−0.47), a state in which food ratings have accumulated ahead of the rest.

The core finding — “how much praise” tracks price, “where the praise lands” does not

The most telling result of the analysis so far is that level and profile behave completely differently with respect to price. The correlation coefficient between the eight-axis average score (how much praise you receive) and estimated settled ADR is +0.61, a clear positive relationship. The correlation between the first principal component (praised for bath and food, or for building and location) and ADR, by contrast, stops at −0.20 — effectively no relationship at all.

Source: HotelBank Editorial Team research (review aggregation) / MetroEngines Research (estimated settled ADR)

What these two figures mean is operationally significant. Accumulating ratings does move price, but the choice of which axis you accumulate them on does not set the ceiling. A prefecture praised for its baths, one praised for its buildings and one praised for its food can all reach the same price band, provided the level is high. How much of the rating level remains unconverted into price does vary by prefecture, and that uncaptured portion is laid out prefecture by prefecture in Listed vs Settled ADR Gap: Japan’s 46-Prefecture Upside Map 2026. Indeed, Bath type Oita sits at ¥15,000, the second-highest ADR level nationally, above Metropolitan type Osaka (¥11,000). Food type Nara, at ¥13,400, also exceeds most of the Hardware type prefectures.

Flip that around and it also means there is no need to force your own prefecture’s or property’s winning formula to match another prefecture’s type. The return on investment for a hot spring property chasing city-hotel-grade location ratings is low; digging deeper into an axis where relative strength is already established is likely the more efficient route to lifting the level.

Source: HotelBank Editorial Team research (review aggregation) / MetroEngines Research (estimated settled ADR)

Where each prefecture’s relative headroom lies

The table below lists all 47 prefectures with their type, eight-axis average score, most relatively strong axis, axis with the most relative headroom, estimated settled ADR and sample base. Note that relative profile values indicate balance within a prefecture and do not imply superiority or inferiority of scores between prefectures. For an example of the same thinking applied at property level, A 4.1 Review Score Is Average, Not a Strength: Sapporo Price Position is a useful reference.

Source: HotelBank Editorial Team research (review aggregation) / MetroEngines Research (estimated settled ADR)

* Covered property count is the property count of the review category with the largest base in each prefecture; review count is the total across the ten categories. The eight-axis average is the simple average of Hospitality, Bathroom, Rooms, Facilities, Breakfast, Cleanliness, Location and Food. Figures in parentheses are relative profile values (in standard deviation units).

A few prefectures stand out. Kyoto shows the strongest relative strength in the country on Cleanliness at +2.49, and recorded the highest national score on five of the eight axes (Rooms, Cleanliness, Location, Hospitality, Facilities). Kochi is relatively strong on Breakfast at +1.60 while carrying headroom on Cleanliness (−2.07). Gunma’s Bathroom score of 4.233 is the highest in the country, with the ratings of Kusatsu, Ikaho and Minakami hot springs translating directly into the prefecture’s character. Ibaraki is relatively strong on Breakfast (+1.12), with headroom on Rooms (−0.96).

The wider market context — rating axes in a phase of flat volume

The practical implications of this structural analysis grow larger in the current market environment, where quantitative growth in accommodation demand has paused. According to the Japan Tourism Agency’s Accommodation Travel Statistics Survey, total guest nights in May 2026 were 54.29 million, down 3.2% year on year, and 46.78 million in June, down 6.5% — both below the prior year. Room occupancy was 61.0% in May 2026 and 56.2% in June. (Note that from the January 2026 survey the stratification basis was changed from employee count to room count, and the survey notes that year-on-year comparisons may include the effect of that change.) For calendar year 2025, annual figures were 661.11 million guest nights, up 0.3% year on year, with room occupancy of 61.6%.

When the tailwind of volume weakens, how you build up rate and satisfaction becomes the main battleground for revenue. What this analysis shows is that there are at least four templates for that build-up, and that the ceiling of the price band you can reach does not differ greatly whichever template you choose. What matters is knowing which template your prefecture or property belongs to and concentrating resources on the axes where relative strength is already established.

Assumptions and limitations of this analysis

Finally, some points to keep in mind when reading the figures in this article.

How much does the level move if you lift one axis? — a conversion based on the definition

So far we have looked at which axis you are praised on. Finally, let us organize how much “how much praise you receive” (the eight-axis average score) moves when you lift that axis. What follows is not a forecast, but a unit conversion derived by rearranging the definition used in this article — that the eight-axis average is the simple average of eight axis scores. No new observed values are used.

Because the eight-axis average is the simple average of eight axes, moving one axis score by +0.05 points changes the eight-axis average by only +0.05 ÷ 8 = +0.006 points. Widening the improvement to +0.15 and +0.25 points yields contributions to the eight-axis average of +0.019 and +0.031 points respectively.

The widest prefectural spread in this article was Bathroom’s 0.534 points. Even if that entire gap were moved on a single axis, the contribution to the eight-axis average would be +0.067 points — reaching only 21.5% of the eight-axis average’s prefectural range (Kyoto 4.298 to Fukuoka 3.987 = 0.311 points). The simple arithmetic confirms it: level does not move on one axis.

Number of axes improved × improvement per axis

At the bottom right of the grid, moving five axes by +0.25 points each yields +0.156 points — still only 50.2% of the 0.311-point prefectural range. Even moving all eight axes by +0.25 points gives +0.250 points, short of the prefectural range. Put the other way round, shifting the level (how much praise you receive) by enough to change a prefecture’s ranking requires simultaneous improvement across multiple axes, whereas the profile (where the praise lands) is determined by the allocation of a single axis. The structure this article identified — that how much praise you receive tracks price while where the praise lands does not — is the flip side of that asymmetry. The implication that digging deeper into an axis where relative strength is already established delivers higher investment efficiency is corroborated by this conversion as well.

Conclusion

Decomposing the guest review baseline of all 47 prefectures into level and profile shows that 52.3% of Japan’s accommodation market is explained by a single axis — prefectures praised for bath and food versus prefectures praised for building and location — and that it can be organized into four types: Bath type with 17 prefectures, Hardware (rooms & facilities) type with 15, Food type with 12 and Metropolitan type with 3.

And while how much praise you receive (the eight-axis average score) correlates with estimated settled ADR at +0.61, the correlation between where the praise lands (the relative profile) and ADR was −0.20, essentially none. Accumulating ratings moves price, but there is wide latitude in which axis you accumulate them on. The clearest proof is that Oita reached a top-tier rate through its baths, Nara through its food, and Kyoto through its buildings and cleanliness. Starting from your own prefecture’s or property’s relative profile and extending the axes where strength is already established is likely the highest-return move available.

Related reading

• Welcome Drinks in Guest Reviews — 255 Hotels Across 47 Prefectures
• A 4.1 Review Score Is Average, Not a Strength: Sapporo Price Position
• Hotel Shuttle Reviews: Airport Shuttle vs Onsen Station Pickup Types
• 1,291 Hotels, 31,024 Rooms Above ¥100,000: Japan’s Luxury Map
• Listed vs Settled ADR Gap: Japan’s 46-Prefecture Upside Map 2026