Japanese keyword research is not English keyword research with a translation step in front of it. The writing system alone breaks the usual method: the same concept can be written three or four legitimate ways, and those variants can carry different volume, different intent and different competitors.
This is the method we actually use, in order.
The problem in one example
Take a common consumer product category. It might be written in kanji, in katakana as a loanword, in hiragana, or in the Latin alphabet. All four are correct Japanese. They are not equivalent as search terms.
- The kanji form often skews formal, older, and toward informational or professional intent.
- The katakana form often skews consumer, modern, and toward commercial intent.
- The hiragana form often appears in casual or mobile queries, and sometimes signals a less expert searcher.
- The Latin form is usually brand-driven.
Google understands that these are related. That does not mean the results are identical, and it certainly does not mean the searcher wants the same thing. Treat script variants as separate keywords until the data tells you otherwise.
This is also where the standard tooling misleads you. Keyword Planner groups variants together and reports one figure. That is convenient and it destroys exactly the signal you need.
Step 1: Start from Japanese, never from your English list
The instinct is to take the English keyword list that works in your home market and translate it. It fails for a specific reason: your English list encodes how your English-speaking customers frame the problem. Japanese customers frame it differently, and often break it at a different point.
Start instead from a Japanese seed — the term a Japanese customer would use for the problem, not for your product — and expand from there with Japanese suggest data.
One caution from our own work: match the tool to the language of the query. Japanese keyword tools are excellent with Japanese input and close to useless with English input. We ran the English phrase “seo in japan” through a leading Japanese tool and got back queries about executive salaries and a Korean actor. Right tool, wrong language.
Step 2: Mine the questions, not just the terms
Japanese search skews long. Queries frequently arrive as full noun phrases or as questions, particularly on mobile and increasingly through AI assistants.
Two sources are worth more than any keyword tool at this stage:
- Yahoo!知恵袋. Japan’s dominant Q&A site. Read the questions in your category and note the exact phrasing, including the misconceptions. The misconceptions are the valuable part — they are the objections your content has to handle.
- Your own sales conversations. What Japanese prospects actually ask, in the words they use. If you have a Japanese distributor or a support inbox, that archive is better keyword data than anything you can buy.
Step 3: Read the Japanese results page before you commit
Volume tells you how many people search. The results page tells you whether you can have any of them.
For each candidate term, look at what Google Japan currently returns and ask three questions.
- What format is winning? If the whole page is ranking articles and comparison media, a product page will not break in. Match the format or pick a different term.
- Who is ranking? If it is entirely large domestic platforms, the term is a brand-awareness target, not an SEO target.
- Is there an AI Overview? If the answer is fully resolved above the results, the clicks are already gone. That term may still be worth targeting — for citation rather than for traffic — but you should decide that deliberately.
This step is where most translated keyword lists die. The term exists, the volume is real, and the page is unwinnable.
Step 4: Check the variants separately
For every term that survives step 3, run the script variants through the check again. You are looking for the case where one variant has lower volume and a dramatically weaker results page. That is where a new entrant gets in.
Also check the spacing. Japanese does not use spaces between words, but Japanese searchers do sometimes insert them between concepts in a multi-part query. Both forms occur, and they are not always treated identically.
Step 5: Group by intent, then decide what to write
Cluster the surviving terms by what the searcher wants, not by string similarity. One page per intent. If two terms deserve the same page, they belong together; if a term deserves a different answer, it needs its own page.
This is the discipline that prevents the most common failure in Japanese content programmes: producing six near-identical articles that compete with each other and collectively rank for nothing. Volume of output is not the goal, and in Japan it is actively penalised as scaled content abuse.
What good output looks like
A usable Japanese keyword plan is not a spreadsheet of 3,000 terms. It is a short list where every row states the term, the script variant chosen and why, the intent, the format that currently wins, and the page that will target it.
If a vendor hands you a large undifferentiated keyword export for Japan, they have run a tool. They have not done the research.
Common questions
Can I just translate my English keyword list?
You can, and it will produce terms that are grammatically correct and commercially wrong. Use it as a checklist of topics, never as a target list.
Which script should we target?
The one your buyers type, verified per term. There is no general rule, though katakana skews consumer and kanji skews formal or professional often enough to be a useful starting hypothesis.
How many keywords do we need to start?
Fewer than you think. A first Japanese content programme that covers ten intents properly beats one that covers a hundred thinly, and it is far more likely to survive its first review.
Working on Japan?
We are a small Japanese SEO practice working with companies entering the Japanese market. Keyword and intent research is done in Japanese, by the people who then write the Japanese. The tooling side is covered in SEO tools for the Japanese market, and what happens after the research in writing Japanese SEO content. More about how we work.
Last reviewed: August 2026.
