Multilingual SEO for Japan: hreflang, Domain Strategy, and Site Structure

Whether Japan is your first foreign-language market or your tenth, two things make it different from the multilingual SEO you may already know: a Japanese-language version cannot be a translation of your English site, and your domain choice is partly a legal question, not just an SEO one. The tags — hreflang, canonical, a domain decision — are the same everywhere. Japan raises the stakes on what goes inside them.

Here is the one-line version. Multilingual SEO for Japan means (1) picking a URL structure Google can geotarget cleanly, (2) implementing hreflang with reciprocal return tags so Google serves the right language, and (3) filling those URLs with content written by native Japanese hands for Japanese search intent — not machine-translated English. Get any one of the three wrong and the other two are wasted.

Those three levers are the backbone of any multilingual SEO program at any number of languages — a multilingual SEO agency running five locales works the same checklist. Japan just adds a legal domain gate most markets lack, so this guide narrows to it. If you are weighing in-house vs. hiring out, the SEO in Japan guide covers build-vs-hire, timelines, and cost; this article stays on the technical decisions — domain, hreflang, canonical, and migration.

One disclosure up front: LOads is itself a bilingual Japan marketing agency, so we have a commercial interest in you hiring a partner like us. We wrote this openly so you can brief an agency competently and sanity-check its recommendations — including ours.

Start with the market reality, not the tags

The audience is there: DataReportal’s Digital 2026 Japan report put internet penetration near 87% (roughly 107 million users at the end of 2025). The practical rule for organic SEO is short — optimize for Google, which also powers Yahoo! JAPAN’s organic results, so ranking on Google largely covers Yahoo! JAPAN too. Do not deliberately block Bingbot either: Bing’s Japan desktop share is higher than most Western marketers expect (inflated by Edge, Windows, and Copilot), though Google dominates on mobile, where most Japanese search happens. We break down why the trackers disagree in search engines in Japan. For consumer categories, discovery also lives partly inside social platforms (social media in Japan); for B2B, Google-first is right.

Domain strategy: the decision that comes before hreflang

Google’s own guidance on managing multi-regional and multilingual sites lays out three URL structures. Everything downstream depends on which you pick.

• Country-code top-level domain (ccTLD) — e.g. example.co.jp or example.jp. Sends the clearest geotargeting signal and the strongest local-trust signal to Japanese buyers, but it is a separate site starting from zero domain authority.
• Subdomain — e.g. ja.example.com. Can be geotargeted in Search Console, but Google treats it as somewhat separate, so authority-sharing is imperfect.
• Subdirectory — e.g. example.com/ja/. Inherits your existing domain’s authority and is simplest to operate (one site, one CMS, one team). This is our default for most foreign B2B companies entering Japan.

Google explicitly lists URL parameters (e.g. example.com?lang=ja) as not recommended for this purpose, so take that option off the table.

A decision framework for foreign companies

The right choice hinges on one question competitors gloss over: are you legally incorporated in Japan yet? That gates the strongest option.

For most foreign B2B companies, especially early in market entry, the subdirectory wins: launch on day one, inherit your English domain’s authority, run one operation. Note the sequencing reality behind the table: a .jp looks low-barrier, but a Japanese postal address itself normally presupposes some local presence — a registered agent, office, or partner — so for a pre-entity foreign company the only genuine day-one option is the subdirectory; both ccTLDs are effectively gated at launch. The ccTLD’s stronger trust is real, but you pay for it with a from-scratch authority climb — worth it once Japan is a committed, resourced market, rarely worth it as a first toe in the water.

You will see agency posts cite precise-sounding stats — “subdirectories rank 27% better,” “63% of B2B sites use subdirectories.” We could not verify those against primary sources, so treat them as marketing claims, not evidence. The durable reasoning is the authority-inheritance logic above, which needs no contested statistic to hold.

The .co.jp legal gate

This is the constraint most guides mention only in passing, and it is a genuine sequencing trap. A .co.jp domain — the one Japanese businesses trust most — is restricted to entities legally incorporated (or registered as a foreign company) in Japan; a company that has not yet incorporated cannot register one. So if you want that trust you must incorporate first — typically as a kabushiki kaisha (K.K.), the standard joint-stock company, or a gōdō kaisha (G.K.), a simpler LLC-equivalent — a months-long process. Japan’s government trade body JETRO (Japan External Trade Organization) runs an Invest Japan support program for foreign entrants. The practical implication: do not put “launch on .co.jp” in a month-one plan with no Japanese entity. Launch on a subdirectory and migrate to .co.jp later if incorporation and trust justify it. Entity choice is its own topic — see starting a business in Japan.

If you already have a Japanese version live

If you run a multilingual site elsewhere, you probably already have a Japanese page — a ?lang=ja parameter, a ja. subdomain, or a translated folder — so your real decision is migrate or leave, not greenfield structure. Leave it if it ranks and the structure is clean; migrate if you are on the parameter URLs Google discourages or you want a subdirectory’s authority-inheritance. The levers: 301-redirect the old URLs (do not delete them), re-map the reciprocal hreflang cluster so no node points at a dead page, and note that Search Console’s change-of-address tool applies to host or domain moves but not to a path-only /ja/ move within the same host. Expect rankings to dip during the transition and recover as Google re-crawls — a botched migration is where most self-managed Japan sites lose ground. That parameter-to-/ja/ move is exactly the kind of project we can walk through against your live setup on request (the pre-consultation below).

Getting hreflang right (and why one broken tag breaks the set)

hreflang is the HTML attribute that tells Google which language/region version of a page to serve to which users. It is where most multilingual sites quietly fail — invisibly, until you notice Google serving your English page to Japanese searchers.

The rules that matter:

• Return tags must be reciprocal. If your English page points to the Japanese page, the Japanese page must point back to the English one. hreflang is a mutual declaration; a one-way link is ignored.
• Every page self-references. Each URL in a cluster lists itself among its hreflang entries.
• Use absolute HTTPS URLs. No relative paths.
• Add an x-default to name the fallback for users who match no listed language.
• One broken node invalidates the cluster. If a single page in the set has a missing or non-reciprocal tag, Google can discard the annotations for the whole group. This is the failure mode to fear — it is not “one page misbehaves,” it is “the set stops working.”

You can declare hreflang three ways, per Google: in the HTML <head>, in an HTTP header (for non-HTML files like PDFs), or in your XML sitemap. For a two- or three-locale site, <head> tags are the most transparent and easiest to audit. To confirm a live cluster actually works, audit it with a crawler like Screaming Frog or a dedicated hreflang validator, and watch Search Console’s Pages/indexing report for return-tag errors — do not assume it is correct because it renders.

hreflang is not canonical — and that resolves the “duplicate content” worry. hreflang tells Google these pages are equivalents for different audiences; it does not merge, dedupe, or canonicalize them. Cross-language pages (your English and Japanese versions) are never duplicates, so hreflang alone handles them. The duplicate risk lives elsewhere: same-language pages across regions (en-US vs en-GB of identical copy) still need a canonical tag — the two solve different problems. For a two-locale en/ja site this rarely bites, but never canonicalize your Japanese page to your English one, which tells Google to drop the Japanese version entirely.

A correct English + Japanese cluster

Here is a copy-pasteable pattern for a subdirectory site with an English page at /en/pricing/ and its Japanese counterpart at /ja/pricing/. Each page carries the identical block:

<link rel="alternate" hreflang="en" href="https://example.com/en/pricing/" />
<link rel="alternate" hreflang="ja" href="https://example.com/ja/pricing/" />
<link rel="alternate" hreflang="x-default" href="https://example.com/en/pricing/" />

Both pages ship the same three lines — that is what makes the return tags reciprocal and gives each page its self-reference.

ja vs ja-JP, and when you need en-JP

Two questions come up constantly:

• ja or ja-JP? Japanese is effectively a single-region language — no “Japanese as spoken elsewhere” variant competes for the same searchers — so plain ja almost always suffices. Use ja-JP only if you have a specific reason to distinguish region, which most sites do not.
• When do you need en-JP? Only if you serve a distinct English page aimed at English speakers in Japan — say, a site for expats or foreign staff that differs from your global English page. If your English page is just your global page, plain en (plus x-default) covers it.

Don’t auto-redirect by language

Google is explicit: do not automatically redirect users to a different language version based on their perceived location or browser language. It can trap Googlebot (which mostly crawls from US IPs) on your English pages so it never sees your Japanese content, and it frustrates users who want the other version. Add clearly visible language links and let the user choose. The Tokyo agency Humble Bunny makes the same point in its Japanese multilingual SEO guidance: give each translated page its own crawlable, indexable URL.

Why “just translate the English site” fails

Even with perfect hreflang, a machine-translated Japanese site underperforms — because SEO is downstream of language, and Japanese is not a language you render by substitution.

Keyword research is not keyword translation. The term a buyer actually types is frequently not the dictionary translation of your English keyword — it may be a shorter loanword, an industry-standard katakana rendering, or a different phrase entirely. Translating your keyword list optimizes for words nobody searches. This has to be done natively.

The same product can live in a different script — and that is a URL and hreflang decision. Written Japanese mixes four scripts (kanji, hiragana, katakana, and romaji), and the same concept written several ways can carry different search volume and intent. For a foreign product this is not academic: a loanword like “cloud” is overwhelmingly searched in its katakana form クラウド, not the English spelling or a kanji translation — so the term that earns the page, the URL you build, and the hreflang node you declare should all track the script your buyers actually type. That choice is made with native eyes, not by translation. We do not invent volume figures here; the mechanics of why one concept fragments across scripts are in Japanese SEO mistakes (the 猫 / ねこ / ネコ demo), and the selection method — pulling volume per variant and picking the winner — is in Japanese keyword research.

Localization means adapting the message, not just the words. In our work with overseas teams, the recurring gap is trust signals: Japanese buyers look for a detailed company profile, a physical Japanese address, executive profiles, and signs of longevity before they convert, and a thin “About” page that works in the US reads as a red flag here. Purchasing is also consensus-driven, often moving through nemawashi (informal groundwork to build agreement before a formal proposal) and ringi (a circulating written approval collecting stakeholder sign-offs) — so your content is read by a committee and must arm an internal advocate to sell you upward. We go deeper in how Japanese B2B buyers decide and website localization for Japan.

What to demand from a Japanese SEO partner

If you are evaluating a vendor, this checklist turns the topics above into buyer leverage. Ask a prospective multilingual SEO agency for Japan to:

• Show a correct hreflang cluster from a live client site — reciprocal return tags, self-references, x-default. If they cannot produce one or explain the “one broken node breaks the set” rule, they do not own this layer.
• State a considered domain-strategy view for your situation, and flag the .co.jp incorporation gate unprompted if you are not yet incorporated. A partner who reflexively pushes .co.jp without asking about your entity status is not thinking about your sequencing.
• Staff native Japanese writers, not translators. Ask to see Japanese-language work samples and who wrote them.
• Do keyword research, not translation — including choosing script (kanji/hiragana/katakana/romaji) per term. Ask how they decide.
• Understand Japanese trust signals and reflect them in on-page content and structured data, not just links.
• Communicate in clear English and give you account and data ownership.

For pricing detail and specific red flags, our Japanese SEO agency guide and choosing a digital marketing agency in Japan go deeper.

Where LOads fits

LOads is a bilingual Japan marketing partner, founded in 2018 in Fukuoka, and we have supported 100+ companies; our CEO, Kosuke Tsunawaki, has 14+ years in digital marketing. The SEO and localization work in this article — domain and hreflang setup, migration, native keyword research and writing — we engage as a fixed monthly fee scoped to the work, not a cut of ad spend. When we also run paid media, that budget is paid from your own accounts, so you keep data and account ownership.

On paid-media cost specifically (our own reference, not an industry standard): that ad-operations fee runs around 20% of media spend, dropping to 15% above ¥10M/month, and clients typically run ¥1M+/month (roughly $6,500+) in media. We also offer a ¥0 spot diagnostic (a view-only read, not the offer below). We share the numbers so “what does a partner cost” is not a black box. For the broader media-cost picture, see what it costs to advertise in Japan.

If you have no Japanese team, the lowest-risk first move is not committing content spend — it is getting a read on the real Japanese SERPs and intent first, so you buy content against evidence, not assumptions. That is the free pre-consultation we offer before any commitment: a read on your Japan market opportunity plus one strategy call, in plain English, at no cost — and the natural place to walk through a domain or migration decision on your live setup. Earlier in evaluating the market? Our B2B marketing market-entry playbook and the broader SEO in Japan guide are good next reads.

Frequently asked questions

Should we use a .co.jp domain, a .jp domain, or a /ja/ subdirectory?

For most foreign B2B companies, a /ja/ subdirectory is the pragmatic default: it inherits your domain’s authority, launches immediately, and runs as one site. A .co.jp carries the strongest local trust but starts from zero authority and is restricted to entities legally incorporated (or registered as a foreign company) in Japan — so a company that has not yet incorporated cannot register one. A .jp is technically open to anyone with a Japanese postal address, but that address usually presupposes local presence, so for a pre-entity foreign company the genuine day-one option is still the subdirectory. Choose on how committed your Japan market is, not the trust signal alone.

Can a foreign company get a .co.jp domain without incorporating in Japan?

No. A .co.jp is restricted to entities legally incorporated (or registered as a foreign company) in Japan. To get one you must incorporate first — typically as a kabushiki kaisha (K.K.) or gōdō kaisha (G.K.). Until then, a /ja/ subdirectory is the workable path.

Do we need hreflang tags if we only have English and Japanese versions?

Yes. hreflang tells Google to serve your Japanese page to Japanese searchers and your English page to everyone else; without it, Google may show the wrong version or treat the two as competing duplicates. Even a two-language site needs a complete cluster: reciprocal return tags, a self-reference on each page, and an x-default. The number of languages only changes how many entries each cluster has.

Should we target ja-JP or just ja, and when do we need en-JP?

Plain ja almost always suffices, because Japanese is effectively a single-region language with no competing national variant vying for the same searchers. Use ja-JP only if you have a specific reason to distinguish region. You need en-JP only when you serve a distinct English page aimed at English speakers in Japan that differs from your global English page; otherwise plain en plus an x-default covers your English audience.

Is optimizing for Google enough in Japan, or do we need to worry about Yahoo! JAPAN and Bing?

Optimizing for Google covers most of the market and Yahoo! JAPAN organic too, since Yahoo! JAPAN’s organic results are powered by Google’s index. Bing is more visible in Japan than Western marketers expect — inflated on desktop by Edge, Windows, and Copilot — so do not deliberately block Bingbot, but you do not build a separate Bing SEO program. Verify current market-share figures before quoting them; they move month to month and depend on device mix.

Why does directly translating our English site underperform in Japanese search?

Because the phrase buyers actually type is often not the dictionary translation of your English keyword, so translating your list optimizes for words nobody searches. Japanese also mixes four scripts, and the same concept in kanji, katakana, or romaji can carry different volumes and intent. Buyers also expect localized trust signals a literal translation cannot supply. Effective Japanese SEO needs native research and native writing, not substitution.

What is the most common hreflang mistake and why does one error break the whole set?

The most common mistake is a missing or non-reciprocal return tag — an English page pointing to its Japanese version without the Japanese version pointing back. Because hreflang is a mutual declaration, a one-way link is ignored. Worse, a single broken node can cause Google to discard the annotations for the entire group, so one error silently disables your whole language-targeting setup.

What should we ask a Japanese SEO agency before hiring them?

Ask them to show a correct hreflang cluster from a live client, state a considered domain-strategy view for your situation, and flag the .co.jp incorporation gate unprompted if you are not yet incorporated. Confirm they staff native Japanese writers, not translators; do keyword research rather than translation (including script choice); and understand Japanese trust signals. Finally, require clear English communication and account and data ownership.