Because the Latin web font a company already licenses for its global site almost never includes full Han (kanji) character coverage, and even when a CJK (Chinese-Japanese-Korean) variant exists from the same type foundry, a global web-font license does not automatically extend to it. The license, the glyph set, and the file weight are three separate problems, and each has to be checked and solved on its own before the Japanese-language site goes live.
Why Can't We Just Reuse Our Global Web Font for the Japanese Site?
A global type system built around a Latin web font almost never ships with the thousands of Han characters Japanese text actually needs, so reusing it for the Japanese-language site produces missing glyphs the moment uncommon characters appear. Most Western-market web fonts are built to cover Latin, and sometimes Cyrillic or Greek, alphabets that run to roughly 100 characters total including punctuation and diacritics. Japanese text draws on thousands of Han characters in everyday use, plus the hiragana and katakana syllabaries, and a font built for a 100-character Latin set was never designed to encode that.
The gap does not show up in a design review, because design reviews use predictable marketing copy: a handful of headlines, a product name, a fixed set of navigation labels. It shows up in production, when the site renders a real person's name, a street address, or a product SKU that happens to contain a character outside the font's limited kanji subset. The browser either shows a visible missing-glyph box or silently falls back to the visitor's default system font mid-page, and either failure breaks brand consistency at the exact moment a real customer's name is on screen. This is the same failure mode covered for printed collateral in Japan Brand Localization: What Foreign Design Teams Get Wrong (2026), but a live website carries a materially higher version of the risk: printed collateral has a fixed, known character set decided before the print run, while a website has to render arbitrary future content, including names and addresses no designer ever previewed.
Does a Global Web Font License Cover the Japanese Version of the Same Font?
No, not automatically. Commercial web-font licensing for CJK typefaces is typically sold and metered on different terms than the Latin fonts most global type systems are built on, and a license purchased for a Latin font from one type foundry does not carry over to a CJK font, even one that is visually paired to sit alongside it. CJK web-font licenses commonly meter usage differently from Latin licenses, for example by monthly pageviews or by a fixed website-license tier rather than a flat one-time fee, because the underlying type foundry's cost structure for producing and maintaining a full Han character set differs from producing a Latin family.
The practical failure mode is a company assuming that because it has cleared commercial web rights for its global brand typeface, a Japanese-language font pulled in to pair with it is covered under the same agreement. It rarely is, particularly if the CJK typeface comes from a different foundry than the Latin one, which is common because relatively few foundries produce both scripts to a professional standard. The license terms, the permitted usage tier, and the renewal or pageview cap need to be checked and secured separately for the CJK font, not assumed to inherit from whatever was signed for the Latin side of the type system.
How Big a Performance Problem Is a Full CJK Web Font?
Large enough to matter: a full Japanese typeface file is significantly heavier than a comparable Latin font file because it has to encode thousands of glyphs instead of roughly 100, and shipping that file unsubsetted can add a meaningful amount to total page weight. Every additional kilobyte of font weight delays first paint on a slow connection, and a Japan-market audience browsing on mobile networks feels that delay directly in load time and, eventually, in bounce rate.
Production Japan-market sites manage this with font subsetting: serving only the glyphs actually used on a given page, or across the site as a whole, rather than shipping the entire character set to every visitor. A subset can be built from the known text on a marketing page, though a subset built this way has to be rebuilt whenever new copy introduces a character outside the original set, which reintroduces the same glyph-gap risk described above if the subsetting process is not revisited on every content update. The other common approach is a webfont loading strategy that prioritizes visible text, showing a system font first and swapping in the licensed CJK face once it loads, so the page is readable immediately rather than blocking on a large font download. Neither approach is a substitute for confirming full Han character coverage in the licensed font itself; subsetting controls file weight for known content, it does not fix a font that was missing the character in the first place.
What Should a Foreign Company Check Before Choosing a CJK Web Font?
Key points:
(a) Confirm the candidate CJK web font's actual Han character coverage against a standard Japanese character set before selecting it, not against the marketing copy on hand at the time; the failure surfaces later, with names and addresses the design team never previewed.
(b) Verify the commercial web-font license for the CJK typeface separately from whatever license covers the Latin global type system, checking the specific metering model (pageviews, website tier, or another basis) and whether it is issued by the same type foundry as the Latin font or a different one.
(c) Plan for font subsetting or a prioritized loading strategy from the start of the build, given that unsubsetted full CJK font files add materially more page weight than a Latin font family, and revisit the subset whenever new site content introduces characters outside the original set.
This same script-pairing discipline extends into typography choices once the licensing and coverage questions are settled: sans-serif Latin fonts pair better with Gothic-style Japanese fonts, serif Latin pairs better with Mincho-style Japanese, and CJK glyphs at small sizes need slightly heavier stroke weight than the paired Latin face to stay legible, particularly for characters with many strokes. Getting the license and glyph coverage right first, then applying pairing and weight rules on top, is the order that avoids redoing the type system twice. Companies running a single Japanese, Simplified Chinese, and Traditional Chinese site off one layout face a compounded version of this problem, covered in How Do You Design One Website for Japanese, Simplified Chinese, and Traditional Chinese Without Three Layouts? (2026), since each script needs its own coverage and licensing check even when the layout is shared.
Frequently Asked Questions
Can we just use a free Japanese web font to avoid the licensing question entirely?
Free CJK fonts often include only a limited subset of Han characters, which reintroduces the exact glyph-coverage gap a commercial license is meant to solve, and free-font terms of use frequently restrict or leave ambiguous the commercial use case a live company website represents. A free font can work for early-stage prototyping, but a production site handling real customer names and addresses needs verified full character coverage and clear commercial usage rights, which is what a paid CJK web-font license is built to provide.
Does font subsetting mean we only pay for a smaller license?
No, subsetting is a performance technique, not a licensing discount; it controls the file weight actually sent to the browser by trimming to the glyphs used on a given page, but the underlying commercial license still needs to cover the full character set the font could ever be asked to render on the live site. A subset built only from today's marketing copy will need to be rebuilt the first time a new name or product term introduces a character outside that subset.
How is this different from the font licensing risk on our printed Japan brochures?
Print collateral uses a fixed, known character set decided at the time of the print run, so a licensing or coverage gap there is caught in proofing before anything ships. A website has to render arbitrary future content, real names, addresses, and user-generated text nobody previewed at design time, so the same font-coverage gap on a website surfaces unpredictably in production rather than in a proof, which is why it needs its own separate check rather than being assumed to follow from print-collateral font clearance.
Conclusion
A global web font and a Japanese web font are two separate licensing and engineering decisions, not one decision extended into a second script. Confirming Han character coverage, securing a CJK-specific commercial license, and building a subsetting or loading strategy around the result are three distinct steps a foreign company's design team has to take deliberately, because none of the three is inherited automatically from the type system already in place for the global site.
This article is informational only and does not constitute legal, tax, or regulatory advice. Consult a qualified advisor before acting on the content. Last updated: August 2026.