Journal
Performance 5 min read

Five Video Ad Formats That Convert, 2026

By Valentina Cruz

Two formats made up most top-performing placements in campaigns run by teams from our 2025 early-access cohort across 14 product categories. Three more served specific roles where alternatives performed less well. These findings were uneven across formats and came from a limited, specific population of early-access teams, mostly in DTC, lifestyle, and consumer product categories creating content for social and connected TV placements, based on 2025 early-access pilot data.

15-second vertical social discovery

In our cohort data, the 15-second 9:16 vertical delivered the strongest return relative to production cost. Completion rates stayed above 65 percent when the first 3 seconds featured a visual surprise or a product moment with strong motion energy. They fell sharply when those first 3 seconds centered on branding, such as a logo, tagline, or color wash. In a 15-second spot, the opening should give viewers a reason to stay, not establish the brand.

This format fits generative workflows because 15 seconds can be planned across 3 to 5 scenes, each with clear visual descriptions. Its length supports quick iteration while leaving enough time for a focused product message when pacing stays tight. In our cohort, teams making 15-second vertical content completed generation, review, and refinement in under 2 hours.

30-second product story, pre-roll and connected TV

For campaigns that needed to establish a specific product narrative, 30-second 16:9 remained the most flexible placement performer. Structure drove the difference: spots using a problem-solution arc, with the problem recognizable in the first 10 seconds, outperformed scene-setting openings by a notable margin in cohort data. This reflects 2025 early-access pilot data from DTC consumer product placements on connected TV and pre-roll.

We should not overstate this finding. Our sample is small, so the preference for this structure cannot be presented as a general rule, even though it aligns with broader industry research on attention patterns in digital pre-roll. The consistent result was narrower: teams whose 30-second scripts introduced a recognizable problem in the first third outperformed teams that began with brand atmosphere.

In generative production, 30 seconds is the point where reference frame quality has the greatest effect. The runtime allows the model to establish and preserve a consistent visual world. Teams that spent time choosing high-quality reference frames before generation saw the quality gap between their outputs and traditionally produced spots narrow significantly.

6-second retargeting bumper

Six-second pre-roll bumpers give generative production a specific structural advantage. They work best as retargeting assets, assuming viewers have seen longer creative and reinforcing one message through one image and one claim. They are quick to produce and fast to iterate.

In traditional production, bumper brevity can make these assets seem low priority, so they are often assigned to junior editors who may not have viewed the complete campaign creative. The result can feel separate from the campaign the bumper should reinforce. With generative production, the bumper uses the full-length spot's reference set and style profile, creating automatic visual consistency. The key creative task is script compression: selecting the longer spot's one moment that can stand alone as a 6-second impression.

1:1 square in mid-feed social placements

Some cohort teams were surprised by square placement results after prioritizing vertical over 1:1. Performance depended on the format and placement: in mid-feed positions on image-heavy social platforms, 1:1 consistently generated the highest view initiation in cohort data, likely because it occupies more screen area than a 9:16 crop in a horizontal-scroll setting.

This is also the format where generating all three aspect ratios in one session helps most. The 1:1 version is not a separate concept, but a center-frame crop from the same visual world as the 9:16 and 16:9 outputs. Teams treating it as a derivative instead of a primary deliverable sometimes failed to compose the visual moment for the square crop. A stronger method is to define that moment in the script for the center-frame square crop, then generate every format from the same input.

90-second brand film for organic reach

The 90-second format is the outlier because its key metric is not click-through. Instead, it is organic share rate and organic search visibility through YouTube indexing. In our cohort, teams creating 90-second brand films with a genuine narrative, rather than extended 30-second spots, saw meaningful organic traffic from search and suggested video placements. Traditional production makes this format easy for small teams to skip: it needs a real shoot day, a scriptwriter able to sustain attention across a longer arc, and more post-production time. Generative production adds roughly 90 minutes to a generation session compared with 30 seconds, changing the effort-to-output ratio dramatically.

One caveat applies: 90-second generative content currently performs best when the brief emphasizes visual atmosphere and brand narrative, not product demonstration. At that length, detailed product close-ups and complex on-screen interactions remain better suited to traditional production. Use generative content for brand story and traditional production for technical product demonstration.

Format is a system, not checklist

These five formats are listed together to show a system, not to say every campaign needs all of them. Decide which placements require coverage, what viewing context each creates, and how the brief should adapt to that context. In our cohort, the strongest aggregate campaign performance came from coherent sets rather than the highest-quality individual spots. Their 6-second, 15-second, and 30-second versions shared one creative foundation and could appear in any order without seeming disconnected.

Generative production makes set-based planning practical for small teams because low per-format overhead can justify producing additional outputs. In our view, this change in format planning matters more than any single performance result enabled by the workflow.

More from the Journal