Every network publishes specs. Those specs are written by engineers describing the system, not by reviewers describing the actual rejection patterns. The result is that most overseas teams get the file dimensions right and still see 30–50% first-pass rejection rates on their Japanese GEO submissions.
This is what we’ve learned from running creative against TrafficJunky and ExoClick specifically for Japan placements. The patterns differ enough between the two networks that a single pre-flight checklist isn’t enough — you need network-specific awareness.
Why first-pass approval matters
Each rejection cycle costs you 1–3 calendar days: a day for the reviewer to respond, time to interpret the rejection note (often vague), iteration time, and a fresh review queue position. If you’re shipping a campaign batch every 2–3 weeks, two rejection cycles consume more than half your launch window.
The first-pass rate isn’t just an efficiency metric — it sets your launch cadence ceiling. Teams that average 90%+ first-pass approval ship roughly twice the creative volume per quarter as teams running 50%.
TrafficJunky’s strictness
TJ runs strict — stricter than its public spec docs suggest. Common rejection triggers we’ve hit on Japanese GEO submissions, in rough order of frequency:
Animated GIF cycle length. TJ documents a maximum GIF duration, but Japanese reviewers in particular flag any GIF whose total cycle feels “rushed” — including ones technically within spec. Sub-2-second cycles get rejected as “disorienting” more often than the spec suggests they should. We target 3-second minimum cycles unless we’re testing a specific high-frequency variant.
Text density on small formats. 300×100 and 320×50 mobile banners with Japanese text get flagged for “illegibility” if any character drops below ~14px display size. Japanese characters don’t compress like Latin — even simple words eat horizontal space. TJ reviewers measure character height, not just font-size declarations.
Border/edge contrast. TJ wants a clear visual edge between the banner and the page background. Light Japanese creative on light page contexts gets bounced for “indistinct boundaries.” A 1px accent border solves most of this; some teams forget this rule because their US/EU placements use darker page contexts.
Logo/lockup placement. Brand mark within the banner needs to sit in one of the two top corners with a minimum margin. Center-positioned logos pass on US placements but get flagged on Japan submissions more often. We don’t have a clean theory why; we just route around it.
ExoClick’s quirks
ExoClick runs looser on most format issues than TJ but stricter on a few specific dimensions:
Compressed file size on animated formats. ExoClick’s stated max file sizes are firmer than TJ’s. Submissions that come in 5–10% under TJ’s limit still get rejected by ExoClick if they don’t optimize. This is mechanical, not editorial — but it catches teams who batch-export to a single set of constraints.
Character rendering edge cases. Half-width katakana (ﾊﾝｶｸ) renders unevenly in ExoClick’s preview pipeline, and reviewers occasionally interpret it as “garbled” and reject. We avoid half-width katakana entirely on creative destined for ExoClick even though it’s technically valid Japanese.
Animation looping spec. ExoClick requires looped animations to clearly indicate the loop boundary — typically a 200–400ms hold on the last frame before restart. Animations that hard-cut from end to start without a hold get flagged as “broken playback.” TJ doesn’t enforce this; ExoClick does.
Format ratio rounding. ExoClick rejects non-exact dimensions. 728×91 (off by one) gets rejected; TJ usually doesn’t notice. This sounds trivial but it’s the most common rejection we see from teams using older export pipelines that introduce rounding errors.
What the docs don’t tell you
Two things that aren’t in either network’s public spec but matter operationally:
Reviewer-to-reviewer variance. On Japanese GEO submissions, who picks up the review affects the outcome more than overseas teams expect. Both networks have a small pool of reviewers for the Japan queue; some are more permissive on edge cases than others. This means a creative that passes one week might get flagged the next. Don’t over-fit your creative to a single reviewer’s threshold — design for the median.
Compliance-by-implication. A small number of policies aren’t documented but get enforced consistently: certain comparative claim patterns, specific imagery treatments, age-suggestion language. The list isn’t long but it isn’t published. Build a relationship with someone on the network’s account-management side if you submit at volume; they’ll usually tell you informally what tripped a rejection.
A pre-flight checklist
The checklist we run against every Japanese GEO submission before sending it to either network:
- Character size: all visible text characters ≥ 14px display size, regardless of font-size declaration.
- GIF cycle: minimum 3-second total cycle on TrafficJunky; loop boundary hold on ExoClick.
- File size: export at 5–10% under each network’s stated max — gives review margin for re-encoding.
- Edge contrast: 1px border or distinct visual boundary on light creative.
- Logo position: brand mark in top-left or top-right with 16px margin minimum.
- No half-width katakana on ExoClick submissions.
- Exact dimensions — re-export anything off by one pixel.
- Animation hard-cut check: preview the loop wraparound; add a hold frame if it cuts harsh.
Running this checklist takes about 10 minutes per batch and has moved our first-pass approval rate above 90% on both networks. The 90% number assumes the content is also clean — the checklist only catches mechanical/formatting issues.
For content compliance (claim language, comparative framing, imagery patterns), that’s a separate post.
If you’ve been getting bounced on Japan submissions and the rejection notes are vague, send us a sample — we’ll tell you which of the patterns above you’re tripping. No pitch attached.