If you're prepping a video for YouTube's in-stream placements, the video ad specs for YouTube in-stream ads boil down to this: use a standard video file (MP4 or MOV), design for 16:9 landscape as the safe default (vertical and square are also accepted in many placements), keep skippable ads built around a 5-second skip window, and keep non-skippable and bumper ads short and tight. The rest is about pacing, captions and sound design so the ad still works once the skip button appears.
The three types of YouTube in-stream ads, briefly
Before you touch specs, know which format you're building for, because the creative rules differ:
- Skippable in-stream ads play before, during or after other videos. Viewers get a skip button, typically after about 5 seconds. There's no strict maximum length, but most advertisers keep these between 15 and 30 seconds for a single-message ad, or longer if they're testing a story-driven format.
- Non-skippable in-stream ads must be watched in full before the content resumes. YouTube caps these to a short runtime (commonly up to around 15 seconds, sometimes slightly more depending on region and current policy), so there's no room for a slow build.
- Bumper ads are the shortest format, built to be watched end-to-end in just a few seconds. They work best as a single idea, a logo, or a tight call to action — not a mini-story.
If you only have one video asset, it's worth cutting it into two or three versions matched to these formats rather than running one length everywhere.
Video ad specs for YouTube in-stream ads: the core technical checklist
These are the specs that actually get checked on upload. Treat the numbers below as the current general guidance — YouTube periodically adjusts thresholds, so confirm the latest limits in its ad specifications page before a big campaign push.
- File format: MP4 or MOV are the standard, most reliably accepted formats.
- Video codec: H.264 is the safe, widely compatible choice.
- Aspect ratio: 16:9 (landscape) is the traditional default for in-stream placements and still the safest choice for broad compatibility. Vertical (9:16) and square (1:1) are increasingly supported too, especially where the ad may also surface in feed-like or mobile-first placements — but always preview before launch, since cropping behaviour can vary by placement.
- Resolution: Aim for a crisp standard HD resolution (1920×1080 for 16:9, or the vertical equivalent) rather than low-resolution footage upscaled after the fact.
- Frame rate: A standard, consistent frame rate (commonly 24–30 fps) plays cleanly across devices.
- Audio: Stereo audio, normalized so it isn't drastically louder or quieter than surrounding content.
- Captions/subtitles: Not a strict upload requirement, but strongly recommended given that many viewers watch with sound off until they decide to engage.
| Format | Typical length | Skip button | Best use |
|---|---|---|---|
| Skippable in-stream | 15–30+ seconds (no hard cap) | Appears around 5 seconds | Story arcs, offers with proof, retargeting |
| Non-skippable in-stream | Short, capped by policy (commonly up to ~15s) | None | Single clear message or offer |
| Bumper | A few seconds, strictly capped | None | Reminder, brand mnemonic, single CTA |
Aspect ratio: what to actually pick
Landscape 16:9 remains the default assumption for classic in-stream inventory, and it's the format least likely to cause letterboxing or cropping surprises. That said, a growing share of YouTube viewing happens on phones held vertically, and the platform has expanded support for vertical and square creative in several placements, including feed-adjacent and Shorts-related inventory.
A practical rule of thumb: if your campaign is specifically targeting classic in-stream (pre-roll/mid-roll on long-form video), build 16:9 first. If you're also running or considering YouTube Shorts or mobile-first placements, have a 9:16 cut ready too, and preview both before spending budget. Many teams now shoot or generate the vertical version first (since it's reusable across TikTok, Reels and Shorts) and then export a 16:9 crop or re-letterbox for traditional in-stream — that's often faster than shooting landscape and trying to reverse-engineer a vertical cut later.
Designing around the skip button
The single biggest creative constraint in skippable in-stream ads isn't a pixel spec — it's the skip button. Everything before that moment has to function as a self-contained mini-ad in case the viewer skips, while still setting up what comes next if they don't.
- Put the product, the problem it solves, or a visually distinct hook in the first 2–3 seconds — don't open on a logo or slow fade-in.
- Avoid a traditional brand intro before the message starts. Lead with the payoff, explain the brand after.
- If the offer or price is the strongest hook, surface it early rather than saving it for the end.
- Treat the 5-second mark as a checkpoint: if someone watched that long, give them a reason to keep watching rather than repeating the hook.
If you're unsure how long your hook should run before it either converts attention or loses it, see what is the ideal hook length for video ads for a format-by-format breakdown. And once ads are live, how to calculate hold rate for video ads is a useful way to see exactly where viewers are tapping skip, so you can tighten the next cut.
Captions, voiceover and sound design
YouTube in-stream ads autoplay with sound in most contexts, which is different from feed placements on other platforms where sound-off viewing is the default to design for. Even so, it's worth building every ad to work on mute too, since viewers can and do mute, and accessibility captions help regardless.
- Add burned-in captions or word-by-word text for the spoken lines, not just a logo and tagline.
- Keep on-screen text large enough to read on a phone screen, since a large share of YouTube viewing happens on mobile.
- If you're deciding between a voiced ad and on-screen talent, this comparison of voiceover vs on-screen talent for video ads walks through the tradeoffs for different budgets and product types.
- Keep music under the voiceover, not fighting with it — a common and avoidable spec-adjacent mistake.
Pre-upload checklist
Run through this before you upload a video ad to YouTube in-stream campaigns:
- File is MP4 or MOV, encoded with H.264.
- Aspect ratio matches the placement you're buying (16:9 for classic in-stream; 9:16 or 1:1 if targeting mobile-first or Shorts-adjacent inventory).
- Resolution is sharp, not a low-res clip stretched to fit.
- Length matches the ad type (longer for skippable, strictly short for non-skippable and bumper).
- Hook or key message lands before the 5-second skip point.
- Captions are present and legible on a small screen.
- Audio is balanced — dialogue clear, music supportive, no clipping.
- You have at least one shorter cut-down ready in case you also want to run bumper or non-skippable versions of the same message.
Where an AI video tool fits into this workflow
Getting the specs right is the easy part; the time sink is usually producing multiple cuts — a full-length skippable version, a trimmed non-skippable version, maybe a vertical cut for Shorts — without starting from scratch every time. This is where FrameNotion can save a lot of back-and-forth: paste a product or landing page link, and FrameNotion AI writes and renders a 30-second vertical ad with a hook, problem, benefit, proof, offer and call to action already structured in, complete with captions and a voiceover. Every ad also exports as 16:9 alongside 4:5 and 1:1, so the same creative can be adapted toward YouTube's landscape in-stream inventory as well as vertical placements like Shorts, without re-shooting anything.
Because FrameNotion generates the ad with distinct structural beats rather than one continuous shot, it's also easier to trim toward a shorter non-skippable or bumper cut later, since the hook and offer sections are already separated. See examples of finished ads, or check pricing — the first ad is free, no card required.
Worth noting clearly: FrameNotion renders the MP4 for you to upload and manage yourself. It doesn't publish directly to YouTube or any ad platform, and it doesn't track in-platform ad performance — you'll still set up the campaign and read results inside YouTube's own tools.
Common mistakes that aren't really about specs
- Using the exact same cut for skippable and non-skippable placements, which forces a rushed non-skippable version that loses its message midway through.
- Treating the first 5 seconds as a branded intro instead of the actual hook.
- Skipping captions because the ad plays with sound on by default — some viewers still mute, and captions improve comprehension even with sound.
- Exporting at a low bitrate to save file size, which shows up as soft, blocky video once it's compressed again by the platform.
FAQs
Frequently asked questions
Can I use the same video for skippable and non-skippable in-stream ads?+
You can reuse footage, but you shouldn't use the identical cut. Non-skippable and bumper ads are capped at much shorter runtimes, so the full-length skippable version needs its own trimmed edit that still lands a complete message.
Does YouTube require captions on in-stream ads?+
Captions aren't a strict upload requirement for most in-stream formats, but they're strongly recommended. Some viewers mute video, and burned-in or platform captions improve comprehension and accessibility either way.
Is there a minimum length for a YouTube in-stream ad?+
There's no meaningful minimum for skippable in-stream ads beyond having enough time to deliver a message, but non-skippable and bumper formats have strict maximums rather than minimums, so the constraint is almost always about keeping the ad short enough, not long enough.
What happens if my aspect ratio doesn't match the placement?+
Most ad platforms will letterbox, crop, or pad the video to fit, which can cut off important text or logos. It's safer to export the correct aspect ratio for the placement you're targeting and preview it before launch rather than relying on automatic cropping.
Do bumper ads need a call to action?+
Yes, but it has to be extremely compact — often a single line of text or a quick verbal cue, since the format is built for a few seconds of attention rather than a full argument.