There is a version of this project that looks finished and is not. The website has been translated into Japanese. A professional agency did it. The Japanese is grammatical, the tone is appropriate, a native speaker signed it off. The pages are indexed. And they get almost no organic traffic.
Nothing went wrong with the translation. The translation did what translation does. It just was not the job.
Translation is faithful to your sentences. SEO has to be faithful to their searches.
A translator’s brief is to carry meaning across a language boundary with as little loss as possible. Given “HR software”, a good translator returns the most natural Japanese equivalent for that phrase in that context. One phrase in, one phrase out. That is correct translation.
Search does not work one-to-one. When a Japanese HR manager, a payroll administrator and a CHRO each go looking for what an English speaker would call “HR software”, they type different things — because in Japanese those are recognised as different product categories, not one category described three ways.
A translator has no way to know that. It is not visible in the source text. The information lives in Japanese search behaviour, not in your English website.
A worked example: one English term, five Japanese markets
Here is what “HR software” actually fragments into. Every one of these is a live commercial query with its own results page, its own competitors and its own buyer.
| Japanese | Romaji | Literal gloss | Who searches it, and for what |
|---|---|---|---|
| 人事システム | jinji shisutemu | “personnel system” | The broad category term. Wide, competitive, mostly informational and comparison content. |
| 人事管理システム | jinji kanri shisutemu | “personnel management system” | Managing employee master data, org structure, placement. Closer to core HRIS. |
| 労務管理システム | rōmu kanri shisutemu | “labour administration management system” | Payroll, attendance, social insurance, statutory labour-law compliance. A different department and a different budget. |
| タレントマネジメント | tarento manejimento | “talent management” | Skills, succession, development. Strategic, executive-level, often driven by a specific initiative. |
| 人事評価システム | jinji hyōka shisutemu | “personnel evaluation system” | Performance reviews, goal setting, appraisal cycles. Frequently bought as a standalone tool. |
The distinction that matters most here is 人事 (jinji) versus 労務 (rōmu). In a Japanese company these are genuinely separate functions. 人事 is the people side — hiring, placement, development, evaluation. 労務 is the administrative and legal side — working hours, payroll, social insurance, compliance with the Labour Standards Act. English collapses both into “HR”. Japanese does not, and neither do the search results.
So if your translator rendered “HR software” as 人事システム — a perfectly defensible choice — your site is now competing in the broadest, most crowded of the five, and is invisible in the four where the buyer intent is sharpest.
Note also that タレントマネジメント is a katakana loanword that stuck, while “HR software” itself did not become ＨＲソフトウェア in general usage. There is no rule that predicts which English terms survive into Japanese and which get replaced by a native compound. It has to be checked, term by term, against actual demand.
The three ways translation loses search demand
One term becomes many. The case above. You pick one, the other four disappear. This is the most common and the most costly.
Many terms become one. The reverse also happens. English marketing vocabulary — “solution”, “platform”, “enablement”, “empower” — often translates into Japanese words that are grammatically fine and commercially empty. ソリューション (soryūshon, “solution”) appears on a great many Japanese vendor sites and is searched by almost nobody as a way of finding a product. A translated page can end up optimised for a word that exists only in marketing copy.
The term exists but nobody uses it. A dictionary-correct Japanese equivalent can be one that practitioners in the industry never say. Every field has this: the textbook word and the working word. Only one of them is typed into a search box.
Intent travels with the word, not with the concept
Even where a translation is the right term, the results page behind it may want something completely different from what your English page provides.
Take a category where the English query “X software” returns vendor product pages, and the Japanese equivalent returns six comparison articles from media companies. Translating your product page does not put you in that result set. The page is competing in a format the results page is not asking for. Working out which is which requires opening the Japanese results and reading them, which is the single most useful hour anyone can spend before commissioning Japanese content.
This is why we do not treat Japanese keyword research as a deliverable that hands over a spreadsheet of terms. The terms are half of it. What ranks for them, and in what format, is the other half.
It changes the page, not just the words
Once the keyword set is built in Japanese, the consequences run through the whole page:
- Which pages exist at all. Five distinct queries with five distinct intents may justify more than one page — or may justify one page and four sections, depending on how much the results overlap. That is a structural decision that has to be made after the research, not inherited from the English site map.
- Titles and headings. A translated title tag inherits English word order and English length conventions. Japanese titles are constructed differently and the natural keyword position is different.
- Depth. Japanese B2B pages tend to run longer and more specific than their Western equivalents. A translated page is exactly as long as its source, which is often short of what the Japanese results page rewards.
- Proof. Case studies, company details and downloadable documents are load-bearing credibility signals in Japanese B2B. They are usually the thing an overseas site has least of.
We go through how this is produced on the Japanese content SEO page. The short form: research in Japanese, brief in Japanese, write in Japanese, and use the English material as source input rather than as a script to be followed.
When translation genuinely is the right answer
It is worth being clear that translation is not the enemy here. Plenty of pages should simply be translated:
- Legal pages, privacy policies and terms.
- Product documentation and support content, where users already know the product and mostly need accuracy.
- Anything whose readers arrive by a route other than search — email, sales follow-up, a link from a partner.
- Pages whose job is to be readable rather than to be found.
The distinction is whether the page has to earn its own traffic. If it does, the words have to come from search demand. If it does not, faithful translation is fine and considerably cheaper.
What to do if you already have a translated site
You do not usually need to start again. A reasonable sequence:
- Build the Japanese keyword set properly, from Japanese sources, without reference to the English site map.
- Map it against the pages you already have. Some will map cleanly; some will map to nothing.
- Rewrite the commercially important pages against the research — usually a small number of pages carrying most of the value.
- Identify the queries that have no page at all and decide which are worth building for.
- Leave the rest translated. Not every page needs to rank.
That is a smaller project than it sounds, and it is almost always cheaper than the translation already paid for.
If you want a view on which of your existing Japanese pages are worth rewriting and which are fine as they are, that is the kind of question our Japanese SEO work starts with — tell us about your site and we will take a look.