Video (VAST)

What it is

Video creatives served through VAST 2.0 / 3.0 / 4.0 tags. The same standard streaming services use. Pre-roll fires before the publisher’s video starts, mid-roll during a programmed break, post-roll at the end. Corner-slider is a non-publisher-video variant — a small player anchored to a content page.

I don’t run video at scale. My Mobidea dataset was push-format. I’m including video for completeness and because push advertisers periodically ask whether to layer video into the mix — so here are the numbers from consulting engagements I’ve run.

What I can show from parallel-buy tests

• eCPMs: Tier-1 EU video pre-roll cleared at $8.40–$14.20 in Q2 2024 (n=420K completed views, four advertiser engagements aggregated). That’s 5–10× push CPM for the same GEO mix.
• VTR (view-through rate): completed-view rate ran 64–78% across the test set. The 22–36% drop-off is where most of the marketing-claim distortion lives.
• CR-given-completed-view: 0.8–1.4%. The CR-given-impression is much lower because not every served impression completes.
• Day-7 stabilisation: video CR holds tighter from day 1 to day 7 than push or interstitial — the consideration was front-loaded in the 15-second creative.

When video is the right format

• Brand campaigns where the message needs 10+ seconds of attention
• VOD and streaming-site placements where the user has opted into a video session
• Sports-content advertisers — high engagement, predictable view completion
• Movie and entertainment offers where the creative carries the story

When it isn’t

• Direct-response performance campaigns under $10K/month budget — video CPMs require scale to converge on a stable bid
• Audiences served only via corner-slider on non-video sites (completion rates drop to 30–40%, the LTV math collapses)
• Verticals where the creative cost exceeds the offer’s available margin

Specs

• Pricing models: CPM
• Devices: Mobile + Desktop
• Geo coverage: 240+ countries
• Anti-fraud: Multi-layer filtering + ML scoring per impression. VAST-event validation (impression, start, q1, midpoint, q3, complete) is a stronger fraud signal than any panel score — wire it in.

For push advertisers: video is a different funnel position. Don’t compare CPMs head-to-head. Compare CR-per-warm-audience-touch across formats over a 14-day window. The picture is rarely what the rate card suggests.