In May 2026, inbound visitors from South Korea reached a record high of 951,300, up +15.2% year-on-year — the highest single-month figure ever, per the Japan Tourism Agency / JNTO preliminary release. But the growth in headcount is not the only signal worth watching. Increased air capacity on Incheon–Kansai and Gimpo–Kansai, earlier vacation windows among university-age travelers, and a shift into what analysts are calling a “routinization phase” of Japan travel are all unfolding simultaneously. In this analysis, we take 28,000 Korean-language guest reviews aggregated by MetroEngines Research (2024–2026) and run them through fine-grained NLP tag analysis to visualize which hotels Korean guests are picking, and why — compared side-by-side with Japanese-language reviewer evaluation axes to surface the overindex gap.

Where the 951K Cluster — 6 Areas Where Korean Guests Concentrate

Of the 951,000 Korean visitors in May 2026, lodging data points to six areas absorbing the bulk of demand: Tokyo, Osaka, Fukuoka, Hokkaido, Okinawa, and Kyoto. Across the ~27,000 hotels MetroEngines Research tracks with confirmed occupancy, aggregating the sample of Korean-language reviews over the trailing 24 months by region yields Hokkaido (N=7,547), Tokyo (N=6,464), and Fukuoka (N=5,836) at the top. Fukuoka’s parity with Tokyo reflects its within-2-hour flight proximity from Busan and Seoul.

May 2026 ADR (Average Daily Rate) is also being pushed up in each area. Kyoto is +24.7% YoY (¥47,300), Hokkaido +10.7% (¥30,400), and Okinawa +9.1% (¥28,100) — the more Korean-concentrated the area, the sharper the price move. Tokyo, by contrast, is only +4.7%: headcount growth alone doesn’t guarantee per-room upside, and the ability to lift ADR varies significantly by region. For a companion read on what’s driving Kyoto / Hokkaido / Okinawa ADR upward, our analysis of long-staying Singaporean demand adds a stereoscopic view: Singapore Inbound Summer 2026 — 7–8-Night Stays and High-ADR Condominium Demand Lifting Kyoto / Hokkaido / Okinawa.

Source: MetroEngines Research, compiled by HotelBank Editorial Team (Korean-language reviews N=28,005, trailing 24-month aggregation)

The Overindex View — What Korean Guests Weigh Differently

Calculating the “overindex” as Korean-guest mention rate divided by Japanese-guest mention rate makes it possible to see, in three dimensions, what repeat-phase Korean visitors are prioritizing when they book a hotel. Taking Tokyo as an example, the location-family tag “Central Access” runs at 20.95% mention rate among Korean guests versus 7.52% among Japanese guests — an overindex of ×2.79. “Tourist-Attraction Access” comes in at ×2.50, and “Shopping Location” exceeds ×12. Meanwhile, “Station Proximity,” “Room Cleanliness,” and “Staff Politeness” — which Japanese guests weight comparably — show mention rates in near-parity, and can be classified as “shared evaluation axes.”

Source: MetroEngines Research NLP analysis, compiled by HotelBank Editorial Team (Tokyo: Korean N=6,464, Japanese N=197,169, trailing 24-month aggregation)

What stands out for Kyoto is that “Japanese Breakfast” hits an overindex of ×15.58. For Japanese guests, washoku breakfast is a default and rarely surfaces in reviews — but for Korean guests, “does this hotel serve a proper Japanese breakfast?” has become a differentiating criterion. Conversely, Osaka’s “Executive Lounge” runs ×12.44 and Hokkaido’s “general Location” hits ×26.48 — meaning the highest-overindex axes shift meaningfully by region. This tells us Korean guests are switching evaluation frames based on the “visit purpose × area” combination.

Area-by-Area Map of Korean-Specific Evaluation Axes

For each of the six areas, we extracted the fine-grained tags with the highest Korean-guest overindex and summarized them in the table below. While location- and central-access-family indexes run high across the board, the region-specific signatures — Kyoto’s “Japanese Breakfast,” Osaka’s “Lounge,” Hokkaido’s “Friendly Service,” Okinawa’s “Room Datedness (negative)” — split clearly.

Source: MetroEngines Research NLP analysis, compiled by HotelBank Editorial Team

The fact that Tokyo, Osaka, and Fukuoka all show “Shopping Location” at ~×12 overindex suggests that repeat-phase Korean guests are increasingly baking “retail access” — rather than “sightseeing access” — into their hotel selection logic. Meanwhile, Hokkaido’s “general Location” overindex of ×26.5 reflects Korean travel patterns that span multiple areas (Niseko, Otaru, etc.), where the “location is good” praise is abstract and cross-area rather than site-specific.

English / Multilingual Support Tag Overindex — How Repeat Korean Guests Rate Language Capability

The hypothesis “Korean guests pick hotels based on Korean-language support” is only partially borne out by the data. On the staff-English-support tag overindex, the top-ranked properties are hotels near Narita and Kansai international airports (Narita Tobu Hotel Airport, Hotel Nikko Narita) and mid-range city hotels in central Tokyo (Hotel Sunroute Plaza Shinjuku, the Kyo-machiya inn Sakura Urushi-tei). On the staff_multilingual (multilingual support) tag, Tokyu Stay Fukuoka Tenjin stands out — showing that in Fukuoka, a short-haul repeat market, “multilingual” beats “just able to communicate the basics” as the evaluated attribute.

Note, however, that the absolute mention rates for staff_english and staff_multilingual tags themselves stay in the low single-digit % range. Korean guests rarely reference “language support” directly in reviews — the evaluation is folded implicitly into general tags like “Location,” “Central Access,” and “Staff Politeness.” In other words, language support has become a necessary condition, no longer a differentiator. For a deeper dive on how popular Tokyo hotels score across languages, see: Tokyo Popular Hotels — Language-by-Language Review Score Gap Analysis.

Source: MetroEngines Research NLP analysis, compiled by HotelBank Editorial Team (top tags across 4 languages, Tokyo, trailing 24-month aggregation)

Placing tag mention rates for the four language groups in Tokyo side by side, the English-speaking and Korean guest evaluation axes track quite closely (weighting Central, Tourist Attraction, and Room Size). Traditional-Chinese speakers (Taiwan / Hong Kong) mirror the Korean pattern with “Shopping Location” also at ~×12 overindex — surfacing a common pattern: “East Asian kanji-sphere guests optimize for retail flow; English-speakers optimize for sightseeing flow.” Only Japanese guests give relatively more weight to “quality for the price.”

Operator-Side Whitespace Opportunities

The data surfaces three operational upsides for capturing the repeat Korean traveler segment.

1. Make the shopping route explicit (Tokyo, Osaka, Fukuoka) — Move “N-minute walk to the main shopping street” ahead of “N-minute walk to the tourist attraction” on hotel-direct and OTA property descriptions. The ×12 overindex means the demand is measurably there; even wording tweaks in plan names and descriptions could unlock upside.

2. Amplify Kyoto’s Japanese-breakfast plans — Kyoto’s washoku-breakfast overindex of ×15.58 signals that “the hotel serves a Japanese breakfast” is itself a differentiation asset for Korean guests. Adding multilingual signaling elements (“Japanese Breakfast,” “Wa no Asa” / Japanese-morning branding) to plan names is worth testing.

3. Lean into Friendly Service in Hokkaido and Okinawa — Hokkaido’s staff_friendly overindex of ×14.10 and Okinawa’s staff_courteous overindex of ×1.53 show that “approachable, warm service” registers cleanly with Korean guests in resort settings. Branding “polite” and “friendly” as separate evaluation axes — rather than collapsing them — is likely effective.