TikTok ad creative best practices: 9:16, 720p, hook

TikTok's basics for performance ads: use sound or music, orient vertically at 9:16, shoot at least 720P, and keep content inside the UI safe zone. Put the hook in the first 6 seconds and the content proposition in the first 3 seconds.

Everything TikTok-side is from Creative best practices for performance ads, a TikTok Ads Manager help page marked "Last updated: June 2025", read 2026-09-29. TikTok calls them general recommendations that have worked for advertisers, not requirements.

What does TikTok recommend for the creative itself?

The page groups advice under a few headings.

How do I generate 9:16 at 720p with Sume?

Sume's video API (POST /v1/videos) has aspect_ratio and resolution fields. 9:16 is a listed aspect ratio, and seedance-2.5 accepts 480p, 720p and 1080p with 4-30 seconds of duration. Each model advertises its own supported ratios and resolutions, so check the model list before you pick one.

The prompt is where the hook lives. TikTok wants the proposition inside 3 seconds, so describe the first shot as the product doing the thing, not an establishing shot.

On variants, TikTok recommends creatives with big differences over near-duplicates, especially when testing. Change the opening shot, the setting or the angle between variants, rather than a single word of the prompt, so the 3-5 clips per ad group are genuinely different.

curl -X POST https://api.sume.com/v1/videos \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: hook-test-01" \
  -d '{
    "model": "seedance-2.5",
    "prompt": "Close-up: water beads roll off a jacket sleeve in the first second, handheld, natural light",
    "resolution": "720p",
    "duration": 8,
    "aspect_ratio": "9:16"
  }'

How do I add text overlays without speech?

Sume's video captions endpoint burns text onto a public video URL. Speech-to-captions needs audible speech, and a silent clip fails as caption_no_speech. For overlay copy, pass cues with text, start and end in seconds; that skips speech-to-text.

Do the math on TikTok's 5-10 words per second before you write the cues: a two-second cue carries roughly 10 to 20 words at that guideline, and a short punchy line fits inside it comfortably.

curl -X POST https://api.sume.com/v1/video-captions \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "video_url": "https://example.com/hook-test-01.mp4",
    "style": "slam",
    "cues": [
      { "text": "Rain-proof in one second", "start": 0, "end": 2 },
      { "text": "Tap to see the fit", "start": 6, "end": 8 }
    ]
  }'

What can Sume not do here?

The safe zone depends on the ad format and TikTok's in-feed ad specs page offers downloadable safe-zone files, so check your text placement against those. The other creative advice, such as featuring real people and trend-led content, is a production choice that no endpoint makes for you.