Serving Visitors to Japan Starts With Not Asking Them to Install an App

Every so often a company asks me to build a dedicated app for visitors to Japan. Reservations, ordering, coupons, guides in several languages. Listed out, the features do look like an app.

From the traveler's side, it looks different. Someone touring Japan on a limited number of days is not going to download an app and create an account for one restaurant or one facility. In my experience, almost nobody does.

I have built products for users in Japan and for users abroad. This is what I would settle first when building a service for visitors.

Visitors in numbers

The Japan National Tourism Organization's final figures for 2025 put visitor arrivals at 42,683,301, up 15.8% on the year before. The top countries and regions:

The detail people miss in this table is Chinese. Taiwan, Hong Kong and Macau add up to about 9.4 million, more than the roughly 9.1 million from mainland China. People in Taiwan and Hong Kong read traditional characters. If your "Chinese version" is simplified Chinese only, it reaches about half of the visitors who read Chinese.

Visitors solve problems with their own tools

The Japan Tourism Agency surveyed 4,110 departing visitors at five airports between November 2025 and January 2026 (survey results).

Asked what gave them trouble during the trip, 15.4% named communicating with staff at facilities and 10.9% named too little or unclear multilingual signage. 43.7% said nothing gave them trouble.

The part I pay attention to is how they dealt with it. Of those who had trouble communicating with staff, 68% used ICT tools to get through it. For signage, 57% did. In other words, most visitors arrive with translation apps and camera translation already on their phones.

So making your pages readable with the tools visitors already carry reaches more people than shipping an app of your own.

Build pages that translation tools can read

In practice:

• Put menus, prices and notices on the page as text, not images. Some tools can read text in images, but browser translation handles real text much more cleanly.
• Stop uploading scanned PDF menus. People end up pinching and zooming on a phone, and translation struggles.
• Write prices as numbers and text, including whether tax is included.
• Keep allergy and dietary information out of small print inside photos.

None of this is a rebuild. It is about how you fix the pages you already have.

A QR code and a web page before an app

If you are building reservations or ordering, I recommend starting with a web page. Scanning the QR code on the table or at the door opens the right page with no login. The language comes from the browser's settings, and visitors can switch it on screen.

Identity checks are where this often breaks. SMS verification sent to a Japanese mobile number does not work for visitors who do not have a Japanese number. Offer email or Apple and Google sign-in, or design the flow so it does not need a login at all.

Forms have the same problem. If furigana, postal code and prefecture are required fields, anyone with an address abroad is stuck. Translate a form built for Japanese users as it is, and this is almost exactly where people stop.

Pick the order of languages from your own data

The national numbers are in the table above, but your customers may not match them. In the Tourism Agency survey, the places where visitors struggled with signage were different in cities and in rural areas.

Start with the data you already have: the browser language of visitors in your site analytics, and, if bookings come through a booking site, the nationality of the people booking. With those two, your first one or two languages usually pick themselves.

You do not need every language at once. Start with English and the one language your data shows most, then add more as you watch inquiries and bookings.

No flags in the language switcher

I often see sites that use flag icons to switch languages. I think it is better to avoid them. A flag stands for a country, not a language. People who read English live in the US, Australia and Southeast Asia, and people who read traditional Chinese live in Taiwan and Hong Kong.

List the languages by their own names: "English," "한국어," "繁體中文," "简体中文." Writing "Korean" in Japanese is no help to someone from Korea.

Also give each language its own URL. If one URL just swaps the text on screen, search engines cannot treat each language as its own page, and people searching from abroad have a harder time finding you. Remembering the language someone picked on the next page is a small thing that helps too.

Where machine translation is enough and where a person should check

Machine translation has become genuinely usable over the last few years. For menu descriptions and facility guides, running it through a machine and fixing what looks off is often enough.

Some parts still need a person: cancellation and refund terms, allergy labeling, age limits and usage restrictions. If the translation is vague there, you cannot point to it later and say "it was written down." This is where a translation budget does the most work.

Check the payment screen too

In the same survey, 10.7% had trouble using credit or debit cards, and 32% of those cases were at restaurants.

If you take bookings or prepayment on the web, test with a card issued abroad to see whether it goes through and whether the card authentication step holds it up. Testing with a Japanese card will not show you this.

At the first meeting

If you are preparing a service for visitors to Japan, tell me three things at the first meeting: which shops or facilities it covers and where they are, how reservations or ordering work today, and whatever customer data you have (site analytics, nationality of people booking and so on).

Thin data is fine. We can start by deciding where to look.