Placement-Native Ad Adaptation From One Master Build

Adaptation requests usually arrive backwards: a finished 16:9 film, a media plan naming nine or ten placements, and a request to make it fit. Cropping that master to 9:16 throws away the composition it was lit for and gives the caption nowhere safe to sit. What comes back is the same ad, smaller and worse. This engagement works from the source instead: your After Effects or Resolve project, the graded ProRes, the separated audio stems. Where the source cannot carry a placement, you are told in week one rather than handed a letterboxed compromise in week three. The direction of work: The smallest and least forgiving placement is built first: muted autoplay, thumb-sized, with a third of the frame owned by platform interface. Everything wider is derived upward from that decision. A build that survives a 1080 by 1920 frame with roughly 480 pixels of TikTok interface across the bottom and 140 down the right will survive a 16:9 pre-roll. The reverse is not true, which is why hero-first adaptation keeps producing assets that technically exist and functionally do not. What gets rebuilt rather than scaled: Framing is re-decided shot by shot, using the 4K or 6K original where there is one, so a two-shot becomes a single rather than two half-faces. Type is re-set at each ratio, not scaled, because a lower third designed for 16:9 and shrunk is unreadable on a phone held at arm's length. Audio is remixed twice: a web mix that reads correctly against the roughly -14 LUFS normalisation the platforms apply, and a separate -24 LKFS pass for any CTV or broadcast cut. Captions are built as a data layer rather than burned pixels, so the next language does not mean a re-render of everything. Formats you receive: H.264 High profile for social and display, ProRes 422 HQ for the CTV partners that require it, VP9 where file weight decides delivery, and the editable .mogrt or Essential Graphics package so your in-house team can produce the following round without coming back. For mobile gaming clients, playable HTML5 builds are scoped separately and have to land under the roughly 5 MB single-file ceiling the networks enforce. That constraint changes the storyboard, so it is agreed before the edit rather than discovered after it. Testing discipline: Variants change one thing. A set where the opening frame, the offer and the end card all moved is a set that cannot tell you which of them moved the number. If your analyst cannot read the variable out of the file name, the variant does not ship. Naming follows your existing convention, not one invented here. Not included: Media buying and account access. Original photography or a new shoot. Rescuing legacy SD or heavily compressed footage through upscaling. Per-asset pricing of banner families. Static display sets beyond the hero sizes attached to a video campaign. Work on assets whose source project cannot be produced can be done from a flat file, but the ceiling is stated in writing first, and it is a low one. Who should not book this: Anyone who needs three hundred display units priced per unit. A template production house does that properly and cheaper, and this practice would only be an expensive middleman in front of it. Also anyone hoping adaptation will fix a concept that is not working. Adaptation moves an asset between contexts. It does not make a weak idea land in more of them, and a wider distribution of the same failure is still a failure with a bigger invoice attached.