Localising creative for Japan and Korea

A creative that performs well in the US and UK often lands flat in Japan and Korea, and the usual first fix, translating the on-screen text, rarely closes the gap. The hooks themselves rest on assumptions that don't travel: that a bold and confrontational voice sells, and that a fast cut rate reads as exciting rather than chaotic. Western briefs likewise treat a generic protagonist as a neutral choice rather than a missed opportunity.

Character does more work than plot

Japanese and Korean mobile audiences respond strongly to a distinct, recognisable character with a defined personality, even in a thirty-second ad for a puzzle or strategy game where the character has no gameplay function. A named and styled mascot or hero figure gives the viewer something to recognise across a campaign's lifespan, the way a long-running gacha or RPG franchise builds loyalty to its cast rather than to a genre. Western creative frequently treats the player-avatar or narrator as interchangeable, a stand-in for "you". That undersells the market's appetite for a character worth following.

This doesn't mean licensing an existing IP. It means giving the studio's own creative a consistent visual identity across concepts, with an expression range and a voice that hold from one ad to the next, rather than treating each ad as a standalone execution.

Text density and pacing

Western hook structure leans on speed: three cuts in the first two seconds, minimal on-screen text, a single word of copy at most. Japanese creative in particular tolerates denser on-screen text and a slower reveal, and often rewards both. A viewer accustomed to text-heavy mobile UI and dense manga panelling doesn't read a screen full of characters as cluttered the way a Western viewer might. Korean audiences sit somewhere between the two, tolerant of text but still responsive to a strong visual hook in the first beat.

The practical implication is that a direct word-for-word localisation of a Western script, dropped into the same cut length, under-uses the available reading time and often looks sparse rather than clean. Extend the caption duration. Add a supporting line of context, or restructure the pacing around a slower first beat; either tends to outperform a literal translation of a Western edit.

Narrator tone matters more than casting alone. A brash, hard-sell voiceover common in US direct-response creative tends to read as aggressive or low-trust in Japan; there a softer and more earnest tone, even a slightly self-deprecating one, performs better for the same underlying claim. Korean creative sits closer to the confident, high-energy register familiar from Western hybrid-casual ads, but still expects a level of polish and production value that a rougher UGC-style clip doesn't carry in that market the way it does on Meta or TikTok in the West.

Colour palettes carry different signals across the two markets. Bright, saturated palettes with strong primary colours read as premium and playful in Korea, closer to how Western hybrid-casual creative already looks. Japan's mobile game advertising leans toward softer, pastel-adjacent palettes for casual and puzzle genres, with saturated colour reserved for action and gacha titles where intensity is part of the pitch. Applying a Western hybrid-casual palette unchanged into a Japanese casual campaign often reads as generic rather than premium.

Cultural references worth testing

References that land reliably across both markets include seasonal motifs: cherry blossom and New Year iconography in Japan, and in Korea a recognisable domestic celebrity aesthetic or K-drama visual grammar (used carefully, without implying an endorsement that doesn't exist). References that reliably fail: American sports metaphors and US-specific humour formats. Irony-led hooks that depend on a Western audience's familiarity with a meme format fail too.

Where to start testing

For a studio with a working Western creative slate and limited local production budget, the sequence that produces the fastest read is to re-cut existing top performers with extended pacing and denser captions before reshooting anything, then test a single consistent character across two or three concepts rather than one-off casting, and only after that run a palette variant test before investing in a full local voiceover pass. Voice localisation is the most expensive lever. Spend on it last, once a cheaper re-edit has already validated the pacing and character choices.

None of this requires abandoning a global creative system in favour of two entirely separate production pipelines. Most studios that get this right keep a shared hook logic and a shared brand identity across markets, then localise pacing and density as distinct, budgeted line items, with palette and voice budgeted the same way, rather than folding them into a single translation pass handled at the end of the process. Treating Japan and Korea as one adjustment rather than two also causes avoidable underperformance; the two markets diverge on pacing tolerance and palette expectations, and on reference points, more than their geographic proximity suggests.

Related archive reading

These articles provide related context and remain subject to their stated review status.

• TikTok for games: beyond Spark Ads
• Why the first playable level should not be level one