Start with the actual placement
Meta Feed and Reels do not share the same viewing surface. YouTube Shorts and in-stream ads are different formats. A platform label alone does not specify the creative constraints.
Before editing, record the intended placement, objective, audience, format and destination. Confirm current requirements in the platform’s official guide rather than relying on a universal “best length.”
A placement review sheet
| Surface | Review in the exported ad | Reference |
|---|---|---|
| Meta Feed or Reels | Crop, text and product visibility, placement-specific safe area and readable offer | Meta Ads Guide |
| TikTok | Vertical composition, sound, hook and message visibility around interface controls | TikTok creative guidance |
| YouTube in-stream or Shorts | Format eligibility, opening, brand integration and duration | Google video ad formats |
The guide links are the specifications source. The checks here are an editorial review method, not a promise that following them produces higher performance.
Adapt one concept without changing the experiment
Consider a hypothetical refill-product demonstration. In each placement the offer remains the same. The adaptation changes framing, caption placement and pacing so the demonstration can be understood in that surface.
Use a three-column brief: what remains constant, what changes for the placement, and what needs checking in playback. If the TikTok cut also changes price and audience, its result cannot isolate the effect of the edit alone.
Inspect the first and last moments
Does the opening establish what is happening without relying on the reviewer’s knowledge of the product? Is the brand recognizable? Does the final ask remain readable at normal playback speed? If a condition is essential to the offer, do not remove it simply to make a shorter cut.
Check sound-on and sound-off comprehension where relevant. Captions can carry spoken information, but they need enough space and time to be read. A still screenshot cannot establish that timing.
Interpret diagnostics within their scope
Use RoastIQ to inspect each cut with the correct context. The five KPIs remain model diagnostics; a platform-native attribute model’s validation is a separate record. A higher score in a different placement is not proof that the placement or cut will deliver better business results.
BuyerLens can explore simulated interpretations when a genuine audience question remains. It cannot establish how a national or platform population will respond.
Decide what to run and back
Keep a record of the versions, review rationale and any remaining uncertainty. Where possible, use comparable campaign evidence to decide which version to invest in. Read the video-length comparison guide for a structured way to compare shorter and longer cuts, or watch SaliencyLab.