SEO for Japanese Websites: Five Checks Only Japan Needs

A Japanese website can look spotless in an ordinary SEO audit and still lose the visitor in the first ten seconds. The price says 税抜 where shoppers expect 税込. A “顧客満足度No.1” badge has no ※ note under it. The kanji look slightly off because the page tells the browser it is written in English. None of that is a broken link or a missing title, so most tools never mention it.

Japanese readers are careful readers. They look for the small print, expect the company to explain itself, and notice when something feels 不自然, unnatural. A lot of that carefulness is written into Japanese law and into the habits of Japanese web teams, which is why we built checks for it.

In short
DeepSpider.ai now has 9 checks written only for Japanese websites. This article explains the five newest: the page language, No.1 claims, tax-inclusive prices, the external transmission notice and clinic advertising. It then covers five general checks that matter more on Japanese sites than elsewhere. None of the Japan-specific checks is legal advice, and most of them are review items that don’t change Website Health Score.

Part 1: five checks only Japanese websites need

1. Japanese text on a page that says it is in another language

Japanese, Chinese and Korean share thousands of characters in Unicode. The character 直 is the same code point in all three, but a Japanese reader expects a slightly different shape than a reader in Beijing. Browsers choose the font from the language the page declares, so a Japanese page marked lang="en" or lang="zh" often shows Chinese glyph shapes. It is readable, and it looks foreign. The W3C explains that screen readers use the same declaration to pick a voice, so a Japanese page read with an English voice becomes nonsense.

This happens more than you would think. Themes ship with lang="en-US", a translation plugin sets the language of the original site, or a template copied from a Chinese version keeps lang="zh-CN".

How DeepSpider decides: the page needs at least 200 letters of visible text, at least 15% of them kana and at least 60% kana and kanji together. Kana are the safe signal, because Chinese text has none. Pages with a few Japanese words in a mostly English text are left alone. The fix is to set lang="ja" on the html element, as the W3C recommends, and mark English passages separately. This is the one Japan-specific check that is a scored Warning, because the problem is certain and the fix takes a minute.

2. No.1 claims without a stated basis

Walk through any Japanese station and you will see it: 売上No.1, 満足度No.1, 業界一. On websites the badge usually sits next to a small ※ note naming the survey, the month and the number of people asked. That note is not decoration. In September 2024 the Consumer Affairs Agency published a report on No.1 claims after several orders under the Act against Unjustifiable Premiums and Misleading Representations (景品表示法). It looked at 368 advertisements and asked 1,000 consumers. Its message to advertisers: a No.1 claim needs a fair comparison group, the right respondents and a sound method, and an “image survey” of people who never used the product is not a basis at all.

ECのミカタ summarised it for online shops in one line: show the outline of the survey next to the claim, and check the basis yourself instead of trusting the survey company.

How DeepSpider decides: it looks for No.1, 1位, 日本一 or 業界一 tied to a measure such as sales, satisfaction or users. A bare ranking list, a goal (No.1を目指す) or a page that names a 調査, 出典 or 調べ anywhere is not reported. It cannot judge whether the survey is fair; it only notices that no source appears in the text.

3. Prices without the tax-inclusive total

Since 1 April 2021 prices shown to consumers in Japan must include consumption tax. The temporary exception that allowed tax-exclusive prices ended on 31 March 2021, as the Ministry of Finance notes. A price such as 9,800円（税込10,780円） is fine; 10,000円（税抜） on its own is not. Shoppers know this too, and a page that only shows 税抜 prices now reads as old or as a business-to-business shop.

How DeepSpider decides: it reports pages with prices marked 税抜, 税別 or 本体価格 when the page names no tax-inclusive price at all. Pages aimed at businesses (法人向け, 卸価格, BtoB) are skipped, because the rule covers prices shown to consumers, not quotes and wholesale lists. Prices drawn by JavaScript after the page loads are not seen.

4. Tracking tags without an external transmission notice

Since 16 June 2023 the Telecommunications Business Act has its own rule for tags that send visitor information to other companies: analytics, advertising pixels, chat widgets. The Ministry of Internal Affairs and Communications says website operators covered by the rule must tell visitors what is sent, to whom and why, on the page or on a page easy to reach from it. Many Japanese sites answer with a page called 外部送信ポリシー or 外部送信について, often next to the privacy policy in the footer.

How DeepSpider decides: it reports Japanese pages that load a known tracking or advertising service when neither the page nor a footer link mentions 外部送信, cookies, オプトアウト or the analytics service. Whether a particular site falls under the rule depends on the business, so read this as a reminder to check, not a verdict.

5. Patient testimonials and before/after photos on clinic sites

Japanese medical advertising rules are stricter than most countries’. Under the 医療広告ガイドライン, patient testimonials about treatment are not allowed on a clinic’s website, including reviews copied from review sites and stories written by staff. The Ministry of Health, Labour and Welfare case book, 5th edition from March 2025, collects real examples from its online patrol. Before/after photos are allowed only with the treatment, cost, period, number of sessions and the main risks and side effects shown next to the photo. Monolith Law Office stresses that a link to another page is not enough.

How DeepSpider decides: it reports clinic pages with a heading such as 患者様の声 or 体験談, or with before/after photos, and says whether risks or side effects are named on the same page. Doctor profiles, staff interviews and recruitment pages are not counted. Cosmetic clinics and dentists are where this matters most.

Part 2: five general checks that matter more in Japan

These checks run on every website, but Japanese sites trip over them more often. Each has a reason rooted in how Japanese is written or how Japanese people search.

1. Title and description length counted in characters

English title advice talks about 50 to 60 characters. In Japanese that would be a paragraph. Google does not publish a character limit at all; its title link guide only asks for concise, descriptive titles. Japanese practitioners such as mobinc usually aim for about 28 to 32 full-width characters. DeepSpider recognises Japanese pages and uses Japanese ranges in Title length and Meta Description length: 15 to 35 characters for titles and 50 to 120 for descriptions, both adjustable in Settings. Main content is counted in characters too, because Japanese has no spaces to count words by.

2. Titles wider than Google shows

What Google actually cuts by is width, not characters. A full-width Japanese character is roughly twice as wide as a Latin letter, so 31 Japanese characters take about as much room as 52 English ones. Titles wider than Google shows measures every title the way a search result draws it. At the default of 580 pixels, about 29 Japanese characters fit. A long title is not wrong in itself, but the most important words should come first, because the end may be cut or rewritten.

3. Garbled characters (mojibake)

Japan has its own word for text that turns into symbols: 文字化け. It usually happens when UTF-8 text is read as another encoding somewhere between the database and the browser, after a server move, a CSV import or an old plugin. The page looks fine to the team that wrote it and broken to everyone else. Garbled characters (mojibake) looks for the typical byte patterns in the HTML, so you find the damaged pages before a customer sends a screenshot.

4. Hreflang language code that looks like a country code

The country code for Japan is JP. The language code for Japanese is ja. Google’s hreflang documentation wants the language first, so hreflang="jp" is ignored. CoDigital, a Tokyo agency that specialises in international SEO, lists it as the most common hreflang mistake on Japanese sites. Write ja, or ja-JP when you also target the country.

5. Bingbot blocked in robots.txt

Most Japanese SEO advice is about Google, and Yahoo! JAPAN uses Google’s results. On phones that is still the picture. On computers it has changed: the government’s 2026 white paper on information and communications reports, from StatCounter data for March 2026, Google at 47.5% and Bing at 46.9% of desktop searches in Japan. Many office workers search from Windows and Edge, where Bing is the default. A robots.txt that welcomes Googlebot and shuts out Bingbot, whether through a copied rule or an old block against “unknown” crawlers, now hides the site from almost half of desktop searches. Bingbot blocked in robots.txt reads the file the way Bing does and tells you if that is happening.

How it works in DeepSpider.ai

The nine Japan-specific checks are the five above and four more that you will meet in your own audit. Most of them sit in their own topic, Technical → Japanese websites. Accounts registered in Japan see the topic by default; anyone else can switch it on in Settings → General → Japanese websites. The checks run on every new audit either way, so turning the topic on later shows results from audits already finished.

Each finding comes with How to fix it guidance in plain language and the source it is based on. The legal checks are written to be quiet rather than loud: we would rather miss a borderline page than fill your report with false alarms. They are reminders to check, not legal advice; for a real decision, ask someone who knows Japanese consumer or medical law. If you translate pages into or out of Japanese, our guide to machine translation is a useful next read.

Frequently asked questions

Does lang="ja" change rankings?

Google says it ignores the lang attribute for working out a page’s language. People don’t: wrong glyph shapes and a screen reader in the wrong language affect every visitor, which is why the check exists.

Is a No.1 claim always a problem?

No. It needs a basis you can show. Name the survey, the date and who was asked, close to the claim, and the check stays quiet.

My shop sells to businesses. Do the price rules apply?

The total-price rule covers prices shown to consumers. DeepSpider skips pages that say they are for businesses, but if your site serves both, show the tax-inclusive price on the consumer pages.

The short version

Japanese visitors read the small print, and Japanese law puts a lot of trust signals into it: the ※ note under a No.1, the 税込 price, the 外部送信 page, the risks under a clinic photo. DeepSpider.ai checks those along with the technical details Japanese sites often get wrong, like lang="ja", hreflang="ja" and titles measured at double width. Fix the trust details first; they are what a Japanese visitor notices before any search engine does.

Sources