GOOGLE ADS DEMAND GEN VIDEO SPECS: ACCEPTED VS SERVED WELL

The thing that took me longest to internalise about google ads demand gen video specs is that "accepted" and "served well" are two completely different states, and the upload screen only tells you about the first one. I've uploaded a video, watched it pass every check, seen the ad go to "Approved", and then watched a whole class of placements receive zero impressions because the clip quietly failed a rule the editor never flagged. This is the view from managing live budgets, not a transcription of the spec sheet: the limits that matter for video are the ones that decide whether your clip actually reaches the inventory you're paying for, or just sits there technically valid and under-served.

So I'll skip the parts you can read on Google's page and focus on the gap. There are a handful of things the google demand gen video ad specs hinge on in practice — video length and the in-stream cliff, the four aspect ratios and which placement each one unlocks, the YouTube-hosted upload mechanic, and the way full coverage feeds delivery. If you want the platform background first, the primer on what Demand Gen is sets the stage; this piece is the video asset layer specifically, and exactly where an accepted clip stops short of being a served one.

One thing worth setting straight before the numbers: Demand Gen video serves the same clips across very different surfaces, and each one has its own floor. Your video can run on YouTube in-stream, the YouTube and Discover in-feed units, YouTube Shorts, Gmail and the Display Network — and a clip that's perfect for one of those can be ineligible for another for reasons that have nothing to do with quality. That's the whole reason the video specs feel fussy. They aren't bureaucracy; they're Google's way of checking you've supplied something usable for every place the ad can show. When I read the video row, I'm really asking one question: which placement does this clip unlock, and what does my reach lose if I skip it.

Accepted vs served: the gap the upload screen hides

Here's the distinction that runs through everything below. Acceptance is a file check — right format, within the size ceiling, above the minimum length. The editor runs it the moment you attach the video, and if it passes, the status goes green. Served well is a different thing entirely: it means the clip actually reaches the placements you built the campaign for, and that depends on rules acceptance never tests. A six-second video is accepted but skips in-stream. A single 16:9 landscape clip is accepted but never competes for Shorts. A one-video ad is accepted but gives the system nothing to rotate when a placement wants a different shape. None of those throws an error.

I learned this the expensive way, watching an "Approved" Demand Gen ad spend almost nothing on in-stream for a week before I worked out the clip was nine seconds long. The specs, read properly, are not a compliance checklist you tick once and forget — they're a coverage map. Every video rule corresponds to a placement, and skipping the rule doesn't get you rejected, it gets you quietly excluded. So the rest of this piece walks the video specs the way I actually check them before launch: not "will this be accepted" but "where will this fail to serve, and have I closed that gap".

Video length and the 10-second cliff

Length is where the biggest accepted-but-not-served gap lives. The floor in the google ads demand gen video specs is five seconds — anything shorter is rejected on upload. But here's the rule the editor won't warn you about: a video shorter than 10 seconds won't serve on YouTube in-stream at all. So a six- or seven-second cut is accepted, the ad reads "Approved", and in-stream stays empty while you wonder why a placement you targeted is getting no impressions. That 10-second cliff is the single thing I see people trip over most with Demand Gen video, because nothing in the interface connects the short clip to the missing placement.

The maximum length depends on the placement rather than a single hard cap, and Google's own recommendation is to use videos longer than 15 seconds to maximise reach and effectiveness across formats. My practical rule from live accounts: keep at least one video at 10 seconds or more if in-stream matters to you, and treat 15 seconds as the working floor for a clip meant to carry reach rather than just exist. If you genuinely want a punchy short cut for Shorts and in-feed, fine — but pair it with a longer one so in-stream has something eligible to serve. One clip rarely covers every length requirement, which is the same logic that runs through the broader Demand Gen asset specs: coverage beats a single perfect asset.

Aspect ratios: four shapes, and the one you can't skip

Demand Gen video supports four aspect ratios: 16:9 landscape (recommended 1920x1080 for HD), 1:1 square (1080x1080), 4:5 vertical (1080x1350) and 9:16 vertical (1080x1920). Those are the numbers. The inference that matters is what each one unlocks. The 16:9 covers in-stream and the desktop in-feed; the 9:16 is what serves the full-screen Shorts inventory, which is where Demand Gen leans heaviest on mobile. My default upload is 16:9 plus 9:16 at minimum, so one campaign covers in-feed and Shorts without the system trying to centre-crop a landscape clip into a vertical slot — which mangles the subject every time and reads as a repurposed afterthought, because it is one.

The 9:16 is the asset I never skip, and it's the one people most often leave out. A missing 9:16 doesn't get you rejected; it just means you never compete for Shorts, full stop — the most attention-dense placement in the format goes unserved while the dashboard still says the ad is fine. Two practical notes from real uploads: keep the action and any text inside the safe zone, because Shorts overlays its UI on the edges and will cover anything you push to the margins, and keep captions on, since a large share of vertical impressions render with the sound off by default. Get the vertical built properly and you cover the highest-attention inventory; lean on a cropped horizontal and you'll watch that placement under-deliver while everything reads as "serving".

Upload mechanics, format and file size

The mechanic that catches people first: Demand Gen video lives on YouTube and is referenced by URL — you can't upload an offline file straight into the ad. You search for a video you've already pushed to YouTube, or you paste its URL. If you don't want the clip on your public channel, an unlisted YouTube video works fine in the campaign. The practical consequence is a workflow step that's easy to forget — the video has to be on YouTube and finished processing before you can attach it — so I upload to YouTube first and build the rest of the ad while it processes, rather than discovering on launch day that the asset isn't ready to attach.

The rest of the format rules are blunt. You can attach one to five videos per ad, the accepted format is .MPG (MPEG-2 or MPEG-4), and the file-size ceiling is a generous 256 GB — generous precisely because the file lives on YouTube, not inside the ad, so it isn't the bottleneck that image and logo sizes are. If a video bounces, it's almost never the size; it's the length floor, the format, or a processing state on YouTube's side. And if you genuinely have no video, Google can auto-generate one from your images and final URL or build one in its video tool — a real fallback, but one I treat as an emergency rather than a plan, because the auto-builds meet the specs and get accepted yet underperform a deliberate upload often enough that I default to custom. That's the same instinct I bring to Demand Gen best practices generally: the assets you build on purpose are the ones that carry the campaign, and the automated fallback is there to stop a gap, not to fill the role.

Headlines, description and CTA on video ads

Video ads carry their own text assets, and the limits are hard ones the editor enforces at the ceiling. The headline is 40 characters and is used in the skippable in-stream format. The long headline runs to 90 characters and is the line that shows in the in-feed video format on Discover and YouTube. The description is also 90 characters, the business name is 25 and required, and the call to action is capped hard at 10 characters — which is tighter than people expect and the field where I most often have to cut a word.

The inference that changed how I write these: when a Demand Gen video ad serves on YouTube Shorts, the system treats headline, long headline and description as interchangeable and can use any of them as the ad's description. That kills the old habit of writing one hero line and treating the rest as filler, because on Shorts any of them might be the line the user actually reads. So I write each text asset as if it has to stand alone, the same discipline I'd apply across a Demand Gen campaign generally — the assets you treat as afterthoughts are exactly the ones the system serves when you're not watching.

What I notice actually moves delivery

The lesson underneath all of this: meeting the video specs gets your clip accepted, but covering them fully is what gets it served broadly. Demand Gen can only deliver what it has an eligible asset for, so a thin video pack that technically passes every check still under-delivers — one clip under 10 seconds skips in-stream, a single landscape ratio skips Shorts, and no amount of bid or budget buys back placements the video can't physically fill. Video coverage is the floor of your reach, not a box you tick on upload.

The signal I watch for this is ad strength. An ad group with both orientations supplied, a 10-second-plus clip in the mix, and several videos to rotate consistently shows stronger ad strength, and stronger ad strength means the system has room to push budget toward the combinations that actually convert. I read it the way I read the diagnostics in any review of how a campaign is pacing: not a grade to chase for its own sake, but a proxy for how much inventory the creative can compete for. Varied ratios, a length that clears the in-stream cliff, more than one video in the pack — that's what I've seen move delivery, far more than any single "best" clip.

The practical takeaway

If you remember one thing about google ads demand gen video specs: the upload screen tells you the file is valid, not that it'll serve where you want. Keep at least one video at 10 seconds or more so in-stream stays eligible, supply both 16:9 and 9:16 so in-feed and Shorts are both covered, upload to YouTube early so the asset is ready when you build, and write every text line as if it might stand alone because Shorts treats them as interchangeable. Those are the checks I actually run before a Demand Gen video ad goes live, in that order.

So I no longer read the video spec sheet as a one-time pass/fail. I read it as a coverage map: every ratio I skip and every length rule I ignore is a placement I've handed back before the auction starts, with no error to warn me. Get the coverage right and the system has room to serve you everywhere; get it thin and it faithfully serves your gaps. The full, current figures live in Google's official Demand Gen asset specifications — read that for the exact numbers, then read this for what they cost you when an accepted clip quietly stops short of being served.