At some point a foreign company entering Japan ends up with a Japanese website it did not build. It arrives with an acquisition, with a local subsidiary that hired an agency three years ago, or with a freelancer who has since stopped answering email.
It is usually WordPress. But it is not the WordPress stack you know. The theme, the SEO plugin, the host and the typography are all drawn from a domestic ecosystem that barely overlaps with the English-speaking one, and the assumptions your team brings will be wrong in specific, expensive ways.
Here is what you will actually find, and what breaks first.
The theme will be one you have never heard of
Astra, GeneratePress, Divi and the rest have very little presence on Japanese corporate sites. The themes that do the work here are domestic products, sold and supported in Japanese:
- SWELL — a paid theme that has become close to a default for Japanese business blogs and media. Heavily invested in the block editor.
- Lightning — free, corporate-site oriented, usually paired with its own block plugin. Very common on small and mid-sized company sites.
- Cocoon — free, blog oriented, enormous installed base among individual publishers.
- SANGO, Snow Monkey and others — smaller, but you will meet them.
This matters for two reasons beyond unfamiliarity.
First, these themes ship a lot of opinion. They generate their own heading styles, table of contents, share buttons, breadcrumbs and structured data. When you add a plugin that also emits breadcrumbs or schema, you get duplicates rather than an override. Audit what the theme already does before installing anything that overlaps.
Second, documentation and support are Japanese-only, including the changelogs. A developer who cannot read Japanese will be guessing at what a theme update changed. Budget for that or plan the migration deliberately — do not discover it during an incident.
The SEO plugin is probably not Yoast
Yoast and Rank Math exist in Japan but are not dominant. On SWELL-based sites in particular you will usually find SEO SIMPLE PACK, a deliberately minimal plugin that handles titles, meta descriptions, OGP and little else. All in One SEO also has a substantial share.
The practical consequence lands on whoever automates anything. Every export script, migration tool and bulk-editing workflow in the English-speaking ecosystem assumes Yoast’s field names. They are not there. Meta descriptions live under different post-meta keys, and in the minimal plugins those keys are frequently not exposed to the REST API by default — so an integration that works everywhere else silently writes nothing.
The fix is small once you know: register the meta keys with REST visibility, then confirm a write actually persists by reading the value back. The failure is silent, so verify rather than assume.
A trap we walked into on our own site
This site runs on that stack, and in August 2026 we found it had no working sitemap at all.
The chain was unremarkable and completely invisible from the outside. A sitemap plugin was installed and active. Its “server” setting was left on WordPress, meaning it deferred to WordPress core’s own sitemap rather than generating one itself. Core’s sitemap had been disabled elsewhere in the configuration. So the plugin waited for core, core never emitted anything, and robots.txt advertised a sitemap URL that returned 404.
Nothing in either component reports an error in that state. Each one is behaving correctly on its own.
Switching the plugin to generate the file itself produced a valid sitemap immediately. We submitted it, and a page that had sat at “Discovered — currently not indexed” for eleven days was crawled and indexed within three minutes. Not three days. Three minutes.
Two things are worth taking from that. The plumbing failure was worth more than any amount of content work we could have done in the same week — and it was only visible because someone requested the sitemap URL directly. Fetch it yourself. Do not infer that it exists because a plugin is active.
Hosting is a domestic decision
Japanese sites are overwhelmingly hosted on Japanese providers — Xserver, ConoHa, Sakura Internet, Lolipop and similar. Foreign teams often read this as inertia. It usually is not.
- Support runs in Japanese, which is what the local staff who operate the site actually need.
- Billing fits Japanese corporate process — bank transfer, invoices that satisfy accounting, and convenience-store payment for smaller accounts. An overseas card-only provider can genuinely fail procurement.
- Domain handling, including the paperwork around a
.co.jp, is routine for them and awkward elsewhere.
Latency is the weakest of the arguments, since a CDN solves most of it. The operational fit is the real reason, and it is a reason to leave a working setup alone unless you have something specific to gain. Consolidating a Japanese site onto the global host is a decision to make with the local team, not for them.
Japanese typography breaks Western layouts
This is where a translated design most visibly falls apart, and it is a rendering problem rather than a writing one.
- There are no spaces between words. Japanese wraps at almost any character, so a layout that relied on word boundaries to break sensibly no longer does. Headings in particular can break mid-term and read as nonsense.
- Line-breaking rules are real rules. Japanese typesetting forbids certain characters — small kana, closing brackets, some punctuation — from starting or ending a line. Browsers implement this, but CSS such as
word-break: break-all, often added to force a Western layout to behave, overrides it and produces text that looks wrong to a Japanese reader. - Japanese web fonts are enormous. A Latin font ships a few hundred glyphs; a Japanese one needs thousands. Loading a full Japanese family costs megabytes, and it is a common and avoidable cause of poor loading performance and layout shift. Use subsetting, or use the system font stack, which on Japanese devices is perfectly respectable.
- Text expands and contracts unpredictably. Japanese is often shorter than English in character count but wider per character. Fixed-width buttons and navigation designed against English strings will overflow or look empty.
What to check before you take over a Japanese site
- Request the sitemap URL and the
robots.txtby hand. Confirm the sitemap returns 200, contains the URLs you expect, and carries plausiblelastmodvalues. - Find out who holds the domain. On inherited sites the registrar account is frequently in a former vendor’s name. This is the item most likely to become urgent at the worst moment.
- List the theme and every plugin, with versions and last-updated dates. Abandoned Japanese plugins are a common finding and a security exposure.
- Check whether Search Console and Analytics are verified to a company account rather than to a departed employee or the agency.
- View the rendered page source for duplicate structured data and duplicate breadcrumbs, the usual result of a theme and a plugin both doing the job.
- Test one REST API write before planning any automation on top of the site.
None of this is exotic. It is simply that the failure modes differ from the ones your team has learned to look for, and the diagnostic trail is in Japanese.
Common questions
Should we migrate the Japanese site onto our global CMS?
Eventually, often yes — consistency is worth real money. But not as the first move. Migrating before you understand what the current site ranks for is how companies lose traffic they did not know they had. Establish the baseline first, then migrate with redirects.
Is the Japanese theme hurting our SEO?
Rarely by itself. These themes are maintained by developers who follow Japanese search closely, and they are usually competent on the basics. The duplicate-output problem caused by stacking plugins on top of them is a far more common issue than the themes themselves.
Can we run the Japanese site from our existing host?
Technically yes. Confirm first that the people who will operate it day to day can get support in Japanese and that your finance team can pay the invoice in a form Japanese accounting accepts.
Our developer does not read Japanese. Is that workable?
For routine work, yes. It becomes a problem at theme and plugin updates, where the changelog is the only description of what changed. Pin versions, and have someone who reads Japanese review updates before applying them.
Working on Japan?
Nihongo Search is published by Amanaut, Inc., a small Japanese SEO and content company in Tokyo working with companies entering the Japanese market, and this site runs on the stack described above — including the sitemap fault, which is why it is in here rather than in a case study. The wider sequence is in our Japan market-entry SEO guide, and the structural decision that comes before any of this is in domain structure for Japan. More about how we work.
Last reviewed: August 2026.
