YouTube Shorts Ads: Buying Mechanics, Specs, and What Outperforms Pre-Roll

Before you do anything else, open your last YouTube campaign and pull the "Where ads showed" placement breakdown. If you assumed you were buying Shorts and the report shows 90% of impressions on in-stream pre-roll, you were not running the campaign you thought you were. That gap is the single most expensive misunderstanding in YouTube Shorts ads, and it traces back to one fact: there is no standalone Shorts placement to buy. You buy a campaign type and an aspect ratio, and the system decides how much of your spend reaches the Shorts feed.

This guide covers how to advertise on YouTube Shorts in practical terms: the buying surfaces, the specs Google actually reads, what Shorts ads cost relative to pre-roll, and the creative differences that let a vertical Short return more per dollar than the same budget on a 15-second skippable pre-roll.

How to advertise on YouTube Shorts: the buying surfaces

Shorts is served, not purchased directly. Two campaign types put creative into the Shorts feed, and your asset's aspect ratio is the strongest input that decides whether you land there.

• Demand Gen is the conversion-oriented surface and the one that should own most Shorts spend. It distributes across Shorts, the main YouTube feed, Discover, and Gmail, and it optimizes for actions rather than views. Feed it a 9:16 asset and Shorts becomes a primary surface in the auction. This is also the surface where Google's machine learning, not your manual targeting, allocates impressions.
• Video campaigns (Video reach, Video views) can reach Shorts too, but they are architected around in-stream pre-roll. A vertical asset is eligible for the Shorts slot; a 16:9-only asset has nothing vertical to serve, so the budget flows to pre-roll by default.

Treat aspect ratio as the placement lever. Uploading 9:16 opts you into Shorts and the other vertical surfaces. Uploading only 16:9 sends spend to pre-roll no matter what you intended, because the auction can only place what you gave it. You cannot fence delivery to Shorts-exclusive, and attempts to force it generally raise cost-per-result, because you are shrinking the auction the optimizer is allowed to work with. The more productive move is to supply vertical creative built for the Shorts context, let Demand Gen allocate, and audit the placement report to see what it actually did. The same logic governs how you buy and ship TikTok ads, where the platform owns delivery and your creative is the steering wheel.

YouTube Shorts ad specs

Google reads the file, then assembles the ad unit. Get the asset right and you stay eligible for the vertical surface; get it wrong and you quietly fall back to pre-roll only.

• Aspect ratio: 9:16 vertical for Shorts eligibility. Provide 16:9 and 1:1 as additional assets so Demand Gen can also fill the main feed and square placements from one campaign.
• Resolution: 1080x1920 is the practical target for the vertical asset. Higher is fine; below 720p wide looks soft on full-screen mobile.
• Length: Shorts themselves cap at three minutes, but for ad creative the conversion-effective band is roughly 15 to 30 seconds. Demand Gen accepts up to about 60 seconds; longer assets bleed completion rate on a swipe-driven feed.
• File: MP4 (H.264) or MOV, under the multi-GB campaign limit. Keep audio in the file. Shorts viewers leave sound on more than most feeds, and a silent asset throws away that advantage.
• Headline, description, CTA: In Demand Gen these are separate text fields, not baked into the video. Google composes them into the ad unit and tests combinations, so write the headline and CTA as distinct assets rather than relying on text inside the frame.

Bidding and objective

If you sell something, run Demand Gen on a conversion objective. Start on Maximize conversions, then move to target CPA once the campaign clears roughly 15 to 30 conversions, which is the minimum the optimizer needs to learn a stable pattern. Reserve Video reach for genuine awareness work with a brand film. The recurring waste is a "views" objective that buys cheap watches: views are inexpensive precisely because the system finds the people most willing to watch and least willing to buy.

YouTube Shorts ads cost: what you're actually paying for

Cost on Shorts depends on whether you are paying for attention or for actions, and the two read very differently in the report.

On view-based buying, vertical YouTube inventory commonly lands in a low single-digit cost-per-view range, often a few cents per view, because Shorts watch behavior is cheap to fill. That number looks great and means almost nothing for a direct-response advertiser, because a cheap view is a weak signal of intent. CPM on Shorts also tends to undercut comparable feed inventory, which is part of why native vertical creative can return more per dollar than pre-roll.

For conversion campaigns, the figure that matters is cost-per-conversion, and the gating cost is the learning period. Demand Gen needs that 15-to-30-conversion threshold before its bidding stabilizes, so a campaign that drives only a handful of conversions a week will look expensive and erratic for reasons that have nothing to do with the creative. In planning terms, the real Shorts cost is not the CPM. It is the budget required to reach enough conversions for the optimizer to learn, multiplied by the number of creative variants you need to find a winner. Underfund either and the unit economics look worse than the surface deserves.

YouTube Shorts ads versus pre-roll on equal spend

Both can serve from the same campaign, yet they reward opposite creative instincts, which is why one consistently outperforms the other for direct response.

Skippable in-stream pre-roll plays before a video the viewer chose to watch. The viewer's mindset is to get past you, and the Skip button arms at five seconds. Everyone in that audience knows you are the toll between them and the content they wanted.

Shorts plays inside a feed the viewer is actively browsing for entertainment. There is no Skip button. The exit is a thumb flick, which is faster than a countdown but governed by a different intent. In pre-roll the viewer wants you gone so the real video can start. In Shorts your ad occupies the same slot as organic content, and a well-built one is hard to distinguish from the Short above it.

On matched spend, vertical-native Shorts creative tends to win direct response for three structural reasons:

• No skip-at-five cliff. Pre-roll forces the entire pitch in front of the five-second arming point. Shorts has no countdown, so a hook can develop for a beat before it has to pay off. That single difference changes how you write the open.
• Full-screen and sound-forward. Shorts is full-bleed vertical with audio that viewers frequently leave on. Pre-roll shares the screen with the page and the "video will play after ad" frame, which dilutes attention before your first second.
• Watch-through compounds. Ads that match feed behavior get watched longer, and longer watch-through feeds back into a lower effective CPM. Native creative is literally cheaper to serve.

The caveat is real: pre-roll still wins when the goal is awareness and you have one polished, broadly appealing brand film. For a founder shipping many direct-response variants against a constrained budget, Shorts is the higher-leverage surface, provided the creative is built for it rather than ported from it.

The creative differences that outperform pre-roll

A pre-roll ad and a Shorts ad should not be the same file, even for the same product. Re-uploading a horizontal pre-roll as a Short is the most common and most costly mistake, because the cropped asset carries every "this is a commercial" signal the Shorts feed is trained to flick away.

• Compose vertical from the start. A center-cropped 16:9 leaves dead bars and cuts off the parts of the frame the eye fixes on. Build for 9:16 production rather than cropping toward it.
• Open without announcing the ad. No logo sting, no slow brand reveal. Begin mid-action with a spoken or on-screen claim, and keep a few opener formulas ready to test. The first second decides the swipe.
• Burn in the on-screen text. Shorts runs sound-on more than most feeds, but baked captions still hold muted scrollers and reinforce the claim for everyone else. Render them into the file rather than depending on auto-captions, which display inconsistently across devices.
• Show one thing happening. A Short that demonstrates the product doing one specific job beats a montage of benefits. Specificity is what reads as real footage instead of a render.
• Ask once, out loud, near the end. A single spoken CTA like "link in the description" or "search the brand name" converts better than two competing instructions.

A Shorts-native timing pattern

The most useful insight here is that Shorts has no skip-at-five cliff, so the timing should be built around that freedom rather than against a countdown. Pre-roll structure jams the hook and the pitch into the first five seconds. Shorts lets the hook breathe to the four-second mark before it earns the swipe, which changes the whole sequence.

1. 0-4s, let the hook breathe. Because there is no Skip countdown, you can spend a full beat establishing motion and a real line before pivoting. Open mid-action with something a person would actually say, paired with visible movement. No logo, no static frame.
2. 4-8s, name the problem in the viewer's words. State the friction the way the viewer would phrase it, not the way your brand deck does.
3. 8-22s, demonstrate the mechanism. Screen recording, before and after, or a hand using the thing. This is the proof, and it has to be concrete enough to be falsifiable.
4. 22-27s, show the specific payoff. One quantified or vivid result, not a list.
5. 27-30s, one CTA. A single spoken instruction. A second CTA splits intent and reliably converts worse in testing.

Here is the pattern applied to a project-management tool aimed at small agencies. At 0-4s, a screen recording shows fourteen browser tabs and a Slack badge climbing, with the line "I had four tools and still missed the deadline." At 4-8s: "Every status update lived in a different app, so nothing was ever actually current." At 8-22s, a real cursor drags three tasks into one board and the deadlines auto-roll. At 22-27s: "One board, one source of truth, and we shipped the next launch two days early." At 27-30s: "Link's in the description." It never opens with the brand name or the phrase "best project tool." It names one problem and shows the fix happening on screen.

Reading the placement report so you know what you bought

Since Shorts and pre-roll share campaigns, the placement report is the only reliable record of where the money went. Check it in the first few days, not at the end of the month, while the budget can still be redirected.

• Pull "Where ads showed." Confirm a meaningful share of impressions came from Shorts and not just in-stream. If almost everything is in-stream, your vertical asset is missing, low-quality, or being out-competed.
• Compare cost-per-conversion by surface. If Shorts converts cheaper than in-stream, that is the signal to weight production toward more vertical variants.
• Watch the view-rate trap. A high view rate with no conversions almost always means the campaign is optimizing for cheap watchers. Re-confirm you are on a conversion goal rather than a views goal.
• Track fatigue by week. Shorts creative decays fast in a high-velocity feed. When cost-per-result on a winner climbs week over week, that is creative fatigue rather than a broken account, and the fix is fresh creative.

Why throughput decides Shorts results

The dependable Shorts approach mirrors any feed: ship many variants of one core message, change a single variable per version, and let spend locate the winner. You cannot predict which hook lands, so funding one polished Short is a test with a sample size of one, which is why most founders who "tried Shorts" never actually read a result.

For one offer, a sane starting test set is four to six hooks against the same demonstration, two or three framings (problem-first, result-first, comparison), and one variable changed per variant so a winner teaches you something reusable. That is a dozen-plus creatives for a single product. Producing them by hand or through an agency is slow and costly enough that the typical pattern is to ship one Short, watch it underperform, and conclude the surface does not work. The arithmetic of the learning period makes this worse: with a 15-to-30-conversion threshold per campaign and creative that fatigues weekly, the throughput required to keep a Shorts program learning is higher than most manual production pipelines can sustain.

Where Shorts will frustrate you

Two limitations are specific to buying Shorts through Demand Gen, and both are worth pricing in before you commit budget.

No surgical placement control. Google allocates across vertical surfaces algorithmically. You influence the split with creative and objective, but you do not dictate it, and there is no clean "Shorts only" switch. If your media plan depends on isolating channels for attribution or budget governance, Demand Gen's pooled delivery will fight you. The placement report is your only after-the-fact accounting, and it is descriptive, not prescriptive.

Faster fatigue than search or display. A high-velocity entertainment feed exhausts creative on a weekly cadence, not a quarterly one. Planning a rolling refresh is not optional in the way it might be on a slower channel, which loops directly back to throughput. The strategy that works on Shorts is exactly the one that hurts most if every variant is a manual edit.

FAQ

Can I run ads only on YouTube Shorts?

Not cleanly. There is no Shorts-only placement to buy. You reach Shorts by running Demand Gen or Video campaigns with 9:16 vertical creative, then confirming in the placement report how much spend actually landed there. Forcing Shorts-exclusive delivery typically raises cost-per-result because it shrinks the auction the optimizer can work with.

What do YouTube Shorts ads cost?

On view-based buying, vertical YouTube inventory often runs in a low cost-per-view range of a few cents, with CPMs that tend to undercut comparable feed placements. For conversion campaigns the meaningful cost is cost-per-conversion, and the gating factor is the learning period: Demand Gen needs roughly 15 to 30 conversions before its bidding stabilizes. Budget to clear that threshold across enough variants, not to chase a cheap CPM.

What are the YouTube Shorts ad specs?

Use a 9:16 vertical asset at 1080x1920, MP4 or MOV with audio, ideally 15 to 30 seconds for conversion creative. Add 16:9 and 1:1 versions so Demand Gen can also fill the main feed and square placements. In Demand Gen the headline, description, and CTA are separate text fields Google composes into the unit, so write them as distinct assets rather than baking them into the frame.

How long should a YouTube Shorts ad be?

For conversion-focused ads, 15 to 30 seconds is the practical band: long enough to hook, demonstrate, and ask for the click, short enough to keep completion high. Note that the optimal length shifts by platform. If yours runs longer, cut the slow middle rather than the hook or the CTA.

Keeping a Shorts program learning means feeding it a dozen vertical, captioned variants and refreshing them weekly, which is where most founders stall. Aitachyon generates batches of 9:16 captioned ad variants from one brand scrape, so the throughput the learning period demands stops being the reason a Shorts test never gets read.