Connected TV Commercial Specifications and Production Tiers

What the platforms ask for

Three documents cover most of the streaming inventory a brand buys directly or through a demand-side platform: Google's non-skippable in-stream rules for YouTube on television, Amazon's specification for Streaming TV and Prime Video, and Roku's creative specification. The rows below restate them; the Sources list links them.

Read the specification as a production brief

Each line in a specification is a decision made before the first frame. A 16:9 requirement with letterboxing discouraged means the piece is composed for 16:9 natively, not adapted from a vertical or square master. A constant frame rate means the edit and every generated shot are conformed to one rate before the mix. A minimum bitrate and a codec mean a clean, uncompressed master is kept and the delivery file is an export from it, not a re-encode of an export. Action-safe framing means the brand, the product and any on-screen text sit inside the area every television shows, because overscan on the set is not something the platform corrects. Loudness is the line most often missed: the platforms and the broadcast standards they follow set loudness targets, and a mix measured against the target the delivering platform names is part of delivery, not a courtesy.

Action safe

The central area of the 16:9 frame that every television is expected to display. Anything the viewer must see, including the brand and the product, is composed inside it.

Title safe

A smaller area inside action safe where on-screen text is placed so that no set crops it. The end card, the address and any legal line are composed inside it.

Constant frame rate

A file whose frames arrive at one unvarying interval. Streaming platforms require it; a piece assembled from sources at different rates is conformed to one before export.

Loudness

The measured perceived volume of the whole spot, as distinct from its peak level. Streaming platforms and broadcasters set a target and reject or re-level mixes that miss it; the delivered mix is measured against the target the platform names.

Pillarbox and letterbox

Black bars at the sides or top and bottom of the frame, produced when a piece of another shape is placed in 16:9. Discouraged or rejected by the streaming specifications; the remedy is to compose for 16:9 rather than pad to it.

VAST tag

The industry-standard XML wrapper a video ad travels in when it is served through a demand-side platform rather than uploaded to one publisher. The creative inside it still has to meet each publisher's specification.

No click, so the creative carries the response

On a television there is no button. Amazon's specification says a non-interactive streaming placement must not include a call to action that invites a click; the other platforms' viewers simply have nothing to click. A commercial written for the feed, which ends by pointing at a button that is not there, wastes its last seconds. The response has to live in the piece: the name said and shown, the product in a frame the viewer can picture again, and where a response is wanted, an address or a code that can be read from the sofa and remembered past the next programme. That is also why the connected TV cut of a campaign is usually the fullest telling of the concept, and why it is planned as the master from which shorter and vertical cuts are taken, as set out in YouTube ad formats and the skip point.

Production tiers, without prices

What a connected TV engagement includes depends on what exists already and how many versions the media plan needs. The tiers below describe scope, not cost; the price of any of them depends on the brief and is stated in a written scope before production begins. None of the tiers carries a delivery time, because that too is set per scope.

How Unreal produces a connected TV commercial without a shoot

The stages are the ones a filmed commercial runs, with the frames coming from generation rather than a camera: a brief that names the product, the claim, the lengths and the platforms; an art-direction document that says what every frame should look like; frames generated to that document and reviewed against the product references; an edit conformed to one frame rate; a grade and a loudness-measured mix; a review of logo, label, proportion and material against the approved references; and delivery files exported per platform specification. What generation changes and what it does not is set out stage by stage in AI vs traditional commercial production; the elements of a brief that still need a camera are in what AI video can and cannot do. Synthetic performers, where a spot uses them, are disclosed as synthetic, and no real person's likeness is created without a written rights agreement.

What this guide does not tell you

It does not price any tier; a written scope does that for a specific brief. It does not state a delivery time. It does not say that a streaming commercial will perform, and it publishes no result from any spot: Unreal never guarantees advertising performance. And it does not replace the platforms' documents: the figures are theirs on the review date, and the delivering media partner confirms the current specification at trafficking.