Optimal Video Ad Length by Platform: What the Data Says

You cut one 30-second ad, upload it to every channel, and let the campaign run. Two weeks later TikTok is cheap and clicking, Meta is fine, and YouTube is quietly eating your budget at a CPM that makes no sense. Same ad, same offer, wildly different cost. The variable you didn't control was length, and each platform prices it differently.

Length is not a creative preference. It's a lever on two numbers you actually pay for: CPM, which the algorithm sets partly on how completely people watch, and CTR, which depends on whether you got to the ask before they scrolled. The right length is the one that protects both on a given placement. That number is not the same on a sound-off vertical feed and a lean-back YouTube pre-roll.

Why length moves CPM and CTR at all

Auction platforms reward creative that keeps people on the platform, because retained attention is what they sell. The signal they read most directly is completion rate and average watch time. A 12-second ad that 60% of viewers finish looks healthier than a 45-second ad that 15% finish, even if the longer one delivered more seconds of message.

That feeds back into your cost. Higher completion generally earns cheaper distribution; lower completion gets taxed with a higher effective CPM. So length affects CPM indirectly, through the completion rate it produces.

CTR is the other half. A click can only happen after the viewer hears the ask. On a short ad the CTA lands while most of the audience is still watching. On a long ad, by the time you reach the CTA you've lost the majority, so your clicks are divided by a much larger impression count. Same clicks, worse CTR, because the denominator includes everyone who left at second eight.

The practical takeaway: length doesn't have one optimum. It has an optimum per placement, set by how that placement is watched.

Short-form vertical feeds: TikTok, Reels, Shorts

These three behave alike for ad purposes. Sound-off-by-default, thumb-driven, autoplay, infinite. The viewer's baseline action is to leave, and they decide in roughly the first second whether to grant you more.

The length that holds up here is short. Creators usually find the paid sweet spot lands in the 9 to 21 second range. Below ~8 seconds you often can't fit hook, demonstration, and CTA without it feeling clipped. Above ~25 seconds, completion rate tends to fall off a cliff and CPM climbs to follow it.

TikTok: 9-15 seconds is the safe core; up to ~21 if the demonstration genuinely needs it. The feed punishes slow openers hardest of the three.
Instagram Reels: similar, 9-18 seconds. Meta's delivery is a little more forgiving of a couple extra seconds than TikTok's.
YouTube Shorts: 15-25 seconds tolerated slightly better than the other two, because Shorts viewers arrive a fraction more willing to watch, but the same scroll physics apply.

The mistake here is treating these as "short video." They're not short versions of a long ad; they're a complete ad that happens to be brief: hook in second one, mechanism in the middle, one CTA at the end. If you have to cut a 30-second concept down to fit, cut the setup, don't compress everything evenly.

Meta feed and Stories: two different clocks

Meta isn't one placement, and its placements don't want the same length.

In-feed (Facebook and Instagram feed): the feed is scrollable but a touch more patient than a vertical short. Ads in the 15 to 30 second band generally do well, with 15 seconds a strong default. You have room for a slightly longer setup than on TikTok, but the same rule holds: the longer you go, the more completion you trade away. Past ~30 seconds you're usually paying for seconds most viewers never see.

Stories: full-screen, tap-to-skip, and the skip is brutal. Treat Stories like a vertical short: 10 to 15 seconds, hook immediately, no slow build. A 30-second Story ad is mostly buying impressions for content nobody reaches.

Because Meta auto-distributes across placements, a single mid-length cut (15-20s, captioned) is a reasonable lowest-effort default. But if a placement is spending real money, cut to its clock rather than letting one length serve all of them.

YouTube: the one platform where longer can be correct

YouTube is the exception that catches people who assume "shorter is always cheaper." It's lean-back, sound-on, and largely watched on bigger screens. The viewer chose to watch a video; your ad is the toll. That changes the math.

The format dictates the length:

Bumper ads (non-skippable, 6 seconds): a hard 6-second cap. One idea, one frame of branding, no room for a story. Useful for reach and reminder, not for explanation.
Skippable in-stream (TrueView): the interesting one. The viewer can skip after 5 seconds, and on a cost-per-view setup you're often charged only if they watch past ~30 seconds or click. That economics actually rewards a longer ad: 30 seconds to a few minutes can work, because the people who self-select to keep watching are your warm audience, and you don't pay for the ones who skip. A 15-second skippable ad leaves the format's main advantage on the table.
Non-skippable in-stream (15-20 seconds): you have the full window guaranteed, so use it; ~15 seconds is the practical target.

The first 5 seconds still carry the whole load on skippable inventory, because that's the free preview before the skip button activates. But unlike a vertical feed, length past that point isn't automatically a tax.

LinkedIn: short, because attention is expensive there

LinkedIn CPMs typically run higher than consumer platforms, so every wasted second costs more in absolute terms. The feed is professional, scrollable, and frequently sound-off. The audience is there to skim, not to watch.

Keep it tight: 15 seconds is a strong target, 30 seconds the practical ceiling. B2B buying cycles are long, but the video ad's job is rarely to close in the feed; it's to earn one click or one moment of recognition. Front-load the relevance ("for RevOps teams running…"), make the value legible with sound off, and put the single CTA early enough that skimmers catch it. Because LinkedIn impressions are pricey, over-shooting length costs more here than anywhere else on this list.

The length decision rule and a per-platform cut sheet

Two reusable artifacts. First, a rule for picking length without overthinking it. Second, a sheet you can cut against.

The decision rule

How is this placement watched? Sound-off and scrollable (vertical feeds, LinkedIn, Stories) means short, hook-first. Sound-on and chosen (YouTube in-stream) means you can go longer.
Does the format charge me for skipped seconds? If yes (feeds, where you pay per impression regardless of watch), every second past the CTA is dead weight, so cut tight. If no (YouTube cost-per-view, charged only on a real view), longer is allowed.
Can the message survive being shorter? Default to the short end of the range. Only add seconds when the demonstration genuinely fails without them. Length should be the minimum that lands the hook, the mechanism, and one CTA.
One CTA, placed where most viewers still are. If your CTA lands after the point where half your completion curve has dropped, the ad is too long for that placement regardless of the clock.

The cut sheet

TikTok: 9-15s, vertical 9:16, hook in second one.
Instagram Reels: 9-18s, vertical 9:16.
YouTube Shorts: 15-25s, vertical 9:16.
Meta feed: 15-30s, 1:1 or 4:5; 15s default.
Meta Stories: 10-15s, vertical 9:16.
YouTube skippable in-stream: 30s-2min, 16:9, first 5s do the work.
YouTube bumper: exactly 6s, 16:9, one idea.
LinkedIn: 15s target / 30s max, 16:9 or 1:1.

The pattern across the whole sheet: the more the viewer chose to be there and the more they can hear, the longer you're allowed to run. The more they're scrolling with the sound off, the more brutal the cut.

The trade-off nobody mentions

This advice has a hidden cost. The correct answer is not "make one ad at the right length." It's "make the same ad at several lengths, one per placement." A 12-second TikTok cut, a 20-second Meta-feed cut, a 6-second bumper, a 45-second YouTube version. That's four edits of one concept before you've tested a single hook variant.

Most advertisers don't do this. They ship one 30-second master to everything because re-cutting per placement is slow and expensive by hand, then they conclude a platform "doesn't work" when really they served it the wrong length. The length is rarely the creative problem. The production cost of getting length right per channel is.

FAQ

What is the best length for a video ad?

There isn't one universal best length; it's set by the placement. For scrollable sound-off feeds (TikTok, Reels, Stories, LinkedIn) aim for 9-21 seconds with the hook in the first second. For Meta feed, 15-30 seconds. For YouTube skippable in-stream, longer is fine and sometimes better — 30 seconds to a couple of minutes — because you mostly pay only for views people choose to keep watching.

Are shorter video ads always cheaper?

On feed placements where you pay per impression, shorter usually protects completion rate and therefore CPM. But on YouTube cost-per-view inventory, you're charged only when someone actually watches past the threshold or clicks, so a longer ad to a self-selecting audience can post a lower effective cost than a forced-short one. The economics of the format decide, not the raw seconds.

Should I use the same video ad on every platform?

You can as a starting point, but it leaves money on the table. Each platform watches at a different length and aspect ratio, and a single master cut will be too long for some placements and miscropped on others. The higher-return move is to re-cut one concept per placement against its clock, then test from there.

Cutting one concept into a TikTok-length vertical, a Meta-feed square, and a longer YouTube version is exactly the per-placement work that stops most teams from getting length right. Aitachyon turns a website URL into a finished, captioned video ad in about two minutes and exports it in 9:16, 16:9, and 1:1 for TikTok, Reels, Shorts, Meta, and LinkedIn, so producing a length per channel costs minutes instead of a round of edits. Starter is $29/mo, Pro $79/mo, Agency $299/mo, with a 14-day money-back guarantee if it doesn't fit how you run ads.