What is Ad format?

An ad format is the structural type of the unit: single image, single video, carousel, collection, responsive text, and the platform-specific variations on each.

A format is not a stylistic choice. It determines the aspect ratio and therefore what fits in frame. It determines how much text renders before truncation. It determines whether motion is available. It determines how many things can be shown and in what order. It determines what a person can do with the ad.

All of that is decided before anybody writes a headline. A message that needs to compare three options cannot be delivered in a single static image, and a message that lands in two seconds does not need a fifteen-second video. Choosing the format for the job is the first creative decision and it is frequently made by default.

Single image. One idea, stated immediately. Cheap to produce, fast to test, and the right choice when the proposition is simple enough to land in a glance. Its limit is that it cannot sequence — everything has to be simultaneous.

Video. The only format that can demonstrate, sequence or build. It earns its cost where the proposition needs showing rather than stating, and it wastes it where a static image would have said the same thing faster.

Carousel. Sequence and comparison. Several products, several features, a before and after, a process in steps. It asks for an interaction, which is a filter: people who swipe are more engaged and there are fewer of them.

Collection and catalogue formats. A browsing surface inside the ad, drawing from a product feed. Suited to retail with range, and dependent on feed quality in a way other formats are not.

Responsive text and asset-based formats. The advertiser supplies components and the platform assembles them per auction. Control is traded for coverage, and the quality of the supplied components is what decides the result.

The most common format problem is not choosing the wrong one. It is supplying an asset made for a different one.

A landscape image on a vertical full-screen surface is cropped, and the crop takes the subject out of frame or leaves empty space where attention should be. Text sized for a large in-feed unit becomes unreadable on a small one. A composition balanced for a square does not survive a portrait crop. Safe areas — the parts of the frame a platform's own interface covers — swallow text that was placed without accounting for them.

Platforms offer automatic adaptation, and it is a fallback rather than a solution. It protects against the worst outcome and does not produce a good one.

The consequence is a real cost of broad placement coverage that accounts frequently do not budget for: buying several surfaces means producing several shapes, and an account that buys them all with one shape is running broken ads on most of them.

Formats differ in the attention they are given, and the difference is mostly about what the person was doing.

A full-screen format occupies the whole screen and is generally given more attention than a unit competing with content around it. An in-stream break interrupts something chosen, which earns attention and spends goodwill. A feed unit is passed at speed. A text ad on a results page answers a question somebody just asked, which is a different proposition entirely.

None of these is better. They are different transactions, and creative that ignores which transaction is taking place performs badly regardless of its quality.

Start from the message. What has to be communicated, and does it need to be shown, sequenced, compared or simply stated. The answer usually names the format.

Check what the placements accept. A format unsupported on the surfaces the campaign buys is not an option however well it suits the message.

Budget the production. Several shapes, not one asset. This is where format decisions become real, and it is where they are most often abandoned.

Test formats against each other, not only creatives within a format. The gap between a static and a video treatment of the same idea is frequently larger than the gap between two statics, and it is tested far less often.

Respect the limits rather than fighting them. Text that will truncate should be written to survive truncation. A first frame that will be a thumbnail should work as one. An ad that will be seen without sound should not depend on sound.

Several platforms have moved toward formats where the advertiser supplies components — headlines, descriptions, images, videos, logos — and the platform assembles and tests combinations automatically.

This changes what the advertiser controls. The unit's final arrangement is not theirs; the quality and range of the components is. A set of near-identical headlines gives the assembly nothing to work with, and a set of genuinely different angles gives it real choices.

The discipline that follows is to supply variety of substance rather than variety of phrasing, and to read the component-level reporting where it is offered, since that is where the remaining signal about what works actually lives.

For an account buying several surfaces, the shapes are predictable and worth producing as a set rather than as requests.

A vertical full-screen asset for stories and reels-style placements, with the subject clear of the areas the interface covers.

A square or near-square asset for in-feed placements.

A landscape asset where in-stream video or partner inventory is being bought.

A first frame that works as a still, because it is the thumbnail and because it is what a fast scroll sees.

Captions, always, since sound-off is the default on most feed surfaces.

Text sized to survive the smallest rendering the campaign will reach.

Produced together from one shoot or one edit session, these cost a fraction of what they cost as five separate requests, and they remove the single most common format failure in accounts with broad placement coverage.