- Shorts ads “look nice but do not lead anywhere” is half right, and the reason is documented: Demand Gen has no call asset, so the platform never learns which calls were accepted.
- Video view is the only campaign type with a real Shorts checkbox, and it optimizes for views. Screen creative on it, then run the winner in Demand Gen with a conversion goal. Never report Video view results as call economics.
- Channel controls sit at the ad group. Set them, spend, then read the placement report. The report is the evidence. The checkbox is a steer.
- Cut to the documented constants: a 3 second call-to-action button in Demand Gen, a 5 second floor, a 10 second in-stream lockout, a 50 second overlay and a 60 second feed cap.
- There is no defensible public cost per call for Shorts, and we will not print one. The reason is structural, and it points at the number you build instead.
Quick answers: The two documents · Which instrument · What a Shorts call costs · The timing constants · The landing page · Conversions and budget · What the evidence proves
Ask around the ads forums about Shorts and the verdict is consistent: they look nice, and they usually do not lead anywhere. If you buy phone calls, take that seriously. It is half right, and the reason has nothing to do with your creative.
Demand Gen has no click-to-call asset and no lead form asset. Google lists both as “not supported” (Google), and the full asset specification is Logo, Image, Uploaded Display ads and Video (Google). So a call that starts on Shorts is never attributed to the placement inside the ad platform. The system never learns which calls a buyer accepted, so it cannot optimize toward them. From the outside that feels exactly like traffic that goes nowhere.
The fix is measurement, not a better hook. The thesis and the tracking stack live in the campaign type that gives you no call button. This page is the Shorts execution that guide does not carry: which campaign type to open, what to cut, and where the tap lands.
Two live Google pages disagree on buying Shorts alone
One Google page offers serving “on YouTube Shorts only” as an example. Another says you “will not be able to buy only YouTube Shorts ads with Demand Gen campaigns.” Both are live, our Demand Gen guide already publishes both quotes and rules on them, and this section starts where that ruling stops.
The tell is Google’s own navigation. The Shorts buying page ends with a next-step link reading “Next: How to buy YouTube Shorts ads with Video view campaigns,” routing a reader who wants Shorts-specific buying away from Demand Gen entirely.
So settle it in your own account. Channel controls sit at the ad group level, not the campaign: ad group settings, then Channels, then “Let me choose.” Google lists seven there, and Maps is easy to miss. The selector is not the whole inventory either, and independent coverage records that surface list as having “expanded steadily” through August 2026 (PPC Land). Budget as if the mix is wider than the one you picked, then read the placement report. What served on Shorts is a reported fact. The checkbox is a steer you verify.
Which instrument to open, and in what order
Video view is the only campaign type with a documented Shorts checkbox: you “buy YouTube Shorts ads with Video view campaigns by checking ‘YouTube Shorts’ in the ‘Multi-format ads’ section” (Google). The same page says what you bought. It “will optimize toward getting as many TrueView views as possible with the given budget,” and a TrueView view on Shorts means the user must “watch the video ad play for at least 10 seconds or until the end of the ad, if it’s less than 10 seconds” (Google).
Our Demand Gen guide rules Video view out for calls, and that stands. It has one job left: screening creative on watch-through, never on call economics.
So run them in sequence instead of choosing between them.
- Screen the creative. Run about a week of Video view and judge the cuts on watch-through alone. You are buying views here, and reporting them as call economics is the mistake this whole section exists to prevent.
- Run the winner for calls. Move it into Demand Gen with your accepted-call conversion goal, and accept a wider inventory mix than Shorts alone.
- Read the placement report. A large Shorts share means the control behaved like a boundary and your read is reasonably clean. A small share means it behaved like a preference, so what you measured is the mix, not Shorts. Either way you funded the mix, so plan the budget that way.
One thing changes the edit between the two. The call-to-action button “appears 3 seconds into the ad for Performance Max campaigns, App campaigns, and Demand Gen campaigns and 10 seconds for Video View campaigns and Video Reach campaigns” (Google). Same footage, two different moments at which the viewer can act.
Why nobody can tell you what a Shorts call costs
Search the adjacent cost queries and you will find a cost-per-thousand figure for Shorts on page one. We traced those numbers, and we are not going to match them.
The missing call asset from the opening is only half of it. The other half is that the pay-per-call networks which do know exactly which calls a buyer accepted never publish traffic-source breakdowns. So the platform that knows the placement does not know the acceptance, and the party that knows the acceptance does not publish the placement. Nobody is withholding this number. It is not being produced.
The nearest public thing to a Shorts-segmented US home-services case study reports ad views, no cost of any kind, never mentions a phone call, and carries no publication date (Strike Social). The circulating cost-per-thousand figures trace to a gated document or an unsourced composite, and the quoted Shorts revenue-per-thousand numbers are creator earnings, the other side of the market entirely.
No public source we could find reports a US Shorts-only cost per buyer-accepted insurance or home-services call.
So build the number you can use. A cost per thousand impressions tells you nothing about whether a buyer paid you. A cost per accepted call does, and a higher cost per raw call can still buy the payable outcome more cheaply when the acceptance rate is better. You can only run that arithmetic on your own funnel, and the measurement that produces it is further down this page.
Cut the video to the constants Google documents
Every incumbent on this search is vague about the edit. Google is not.
| Constant | Documented value | What it means for the edit |
|---|---|---|
| Call-to-action button, Demand Gen | 3 seconds in (Google) | Your hook has to land before the button exists |
| Call-to-action button, Video view and Video Reach | 10 seconds in (Google) | Same footage, later ask |
| Minimum video duration | 5 seconds (Google) | The floor |
| In-stream eligibility | Videos under 10 seconds cannot serve on YouTube in-stream (Google) | A 7 second Short works on the feed and locks you out of in-stream |
| Action-oriented Shorts | 10 to 30 second vertical videos (Google) | The working range for a call offer |
| Match user behavior | Add video ads under 60 seconds (Google) | Shorter cuts are the native length on this surface |
| Shorts feed cap | Up to 3 minute videos, but only the first 60 seconds plays on the feed (Google) | Past 60 seconds plays only if the viewer taps through to the watch page |
| Watch-page overlay | An overlay appears at 50 seconds letting the user watch the rest on the YouTube watch page (Google) | Your ask should land before the overlay, not after it |
| Vertical image spec | 9:16, minimum 600×1067, recommended 1080×1920, marked recommended for Shorts (Google) | Shoot and export to this, not to a cropped landscape master |
| 9:16 safe zones | “Safe zones for 9:16 images are similar to 9:16 videos” (Google) | Keep the offer away from the interface controls |
| Static images on Shorts | Square and portrait static assets can serve with an enhanced background (Google) | Selecting the Shorts channel is not the same as committing to video creative |
| Hosting | Videos must be uploaded to YouTube, public or unlisted (Google) | The operations step people forget |
This is where a Shorts cut stops behaving like an in-stream cut. On in-stream the opening is about surviving the skip, which we wrote up in what YouTube builds that last-click never shows. On the Shorts feed there is no skip button to survive. The constants above are the brief instead.
The production template
Our recommendation, not Google’s specification, built to the constants above.
- 1 to 3 seconds. Identify the person, the problem, or the decision. This has to land before the Demand Gen button appears.
- 5 to 10 seconds. One useful reason to care, with visible evidence.
- Middle. Who it is for, and the qualification that matters.
- Close. Exactly what happens after the tap, and who answers.
Screening the same cut on Video view or Video Reach moves one thing: the button does not exist until 10 seconds, so hold the explicit ask a beat longer than the Demand Gen edit does.
Google’s own production guide backs the shape: “Human focus right from the beginning,” “Fast pacing, tight framing, 2+ shots in the first 5 seconds work best,” and “Talent articulates ‘Call-to-Action'” (Think with Google).
Shoot vertical at 1080×1920, with clean audio, readable captions, and important text away from the interface controls. Use your actual agents or technicians when their expertise is the reason to trust you. Trade coverage of Google’s own creative guidance records the same advice, telling advertisers to supply assets “in vertical, square, and landscape formats” and to create content “that feels native to the placement” (Search Engine Land). For a Shorts-first test, vertical leads; square and landscape earn their place only once you accept in-feed and in-stream inventory alongside the feed.
Typed into the platform, that looks like this.
| Element | Private health | Home services |
|---|---|---|
| On-screen hook, first 3 seconds | “Self-employed? The marketplace is not your only option.” | “Told you need a whole new AC system?” |
| Spoken close | “Tell us your age, your state and what you need covered. A licensed agent walks you through what you actually qualify for.” | “We send a second technician for a written second opinion before you spend anything. Give us your ZIP and what the first company told you.” |
| Button text | “Tap to see your options, then speak with an agent” | “Tap for a second opinion, then talk to a technician” |
Keep the button honest about the next screen. Anything implying the button itself places a call is inaccurate, and Demand Gen has no button that does.
Where the tap has to land
Your viewer is holding the phone they would call from. That single fact settles the capture design: a QR code is the wrong primary action here. It is the right answer on a television, and we made that case in how a QR code captures a response from a TV screen. On Shorts it asks someone to scan the screen they are already holding.
So the destination is a fast mobile page with a tap-to-call button reachable without scrolling. Two routes are worth testing, and choosing between them is an economics decision.
Direct call. The offer, the eligibility or service area, the trust information, and a large call button. Lowest friction to an agent or a contractor.
Short qualification. Two or three essential questions, then the call button. Worth the friction only when your rejection rate justifies it. Home services: ZIP and service type. Private health: age, state, product fit.
Skip mandatory long forms and one-time-password steps in a first test unless they solve a problem you have measured. And count a requested callback as its own funnel outcome. Reporting it alongside a consumer-initiated inbound call hides which one you can actually sell.
One honest caveat, subtler than it looks. Google’s QR-code page does list a Shorts placement, where ads “appear between organic Shorts videos” and “The QR code appears on the screen to drive users to the site.” But the same page says QR codes “are exclusively available for Connected TV (CTV)” and are “optimized to appear only when your ad is viewed on a TV” (Google). So a Short watched on a television can carry a QR code. The phone in your viewer’s hand cannot.
Feed the bidding system a call it can learn from
Everything above decides what serves. This decides what the campaign learns from.
The goal map
The conversion action you bid on is the most important setting in the account.
The buyer-accepted call is the one that matches how you get paid, and the lever is a campaign-specific custom goal: “Conversion actions included in a custom goal are used for reporting and bidding regardless of whether they’re set as ‘primary’ or ‘secondary’ actions” (Google). That is why the custom goal matters and the primary or secondary label does not.
Wiring the outcome back into Google
The path is short to say: ad, fast mobile page, tap to call, routed to a buyer, qualified outcome returned to Google Ads. The number insertion, offline import and verification gate are already built out in the campaign type that gives you no call button, the platform choice sits in our roundup of call tracking platforms with a published price, and the setup itself is wiring call outcomes back into Google.
What matters here is lag. Google Ads auto-tagging has to be on, and Ringba then documents upload delays of up to 300 minutes for a call-only ad and 1400 minutes for a click ad (Ringba). Same-day reporting is incomplete by design, so do not judge a test on the day it ran. Enhanced conversions for leads buys accuracy, not a match rate: “more accurate conversion reporting compared to standard offline conversion import” (Google).
One thing to know before you shop: Demand Gen gives a call-tracking platform no native capture to plug into, so whichever one you pick, the route home is the same. Capture the click identifier on your own landing page, then send the outcome back as an offline upload.
Bidding and budget
One paragraph, because we have written this three times already.
Start on Maximize conversions while the target is unknown, then move to Target CPA once your accepted-call action has recorded at least 50 conversions, which is Google’s own guidance (Google); Target CPA works with no history (Google). Set that target near your mature actual cost on the same accepted-call definition, never on a form lead from another channel, and set it deliberately, because after the August 17 change a campaign that was beating its target may now deliver closer to it: see the August 17 bidding plan, why your best campaigns were the ones exposed, and the prep checklist. Leave view-through bidding off: “VTC optimization does not currently support offline conversions, store visits, or OCI” (Google).
Budget follows the cost and frequency of that event. For campaigns using target CPA bidding, Google “recommends setting a budget that’s at least 10 times your target CPA” (Google). Google’s sentence names no period. A Google Ads campaign budget is set as an average daily amount, so read the recommendation against your own budget setting rather than as a price of entry. Learning “can take up to around 50 conversion events or 3 conversion cycles,” aimed at “within 2 weeks” (Google), and a daily budget can spend up to 2 times its average on a given day (Google). Then leave it alone: “Changing settings forces the campaign to do some or all learning over again” (Google). Fund your strongest offer before splitting a budget across two.
The metrics you have to build yourself
This is the benchmark nobody will sell you.
Review click-through, engaged-view and view-through attribution separately, and never add Google and Meta totals together as unique calls, because each platform counts a call it was involved in. And if you ever do turn view-through bidding on, expect Google Ads to diverge from your own records: “impression based metrics like Engaged-view conversions (EVC) and VTC are not exported” (Google). For pricing, what an insurance call costs by screening depth and setting a duration floor against a buyer’s billable threshold; for acceptance, verifying a caller before you pay for the lead.
One policy note that decides whether the campaign can run at all. To advertise health and medical insurance coverage in the US you must be certified by Google, which requires certification with G2 Risk Solutions first (Google). That is G2 Risk Solutions, not Google Guaranteed, which is a Local Services Ads program.
The evidence that exists, and what it does not prove
There is documented evidence that Demand Gen produces leads. There is no documented evidence that Shorts produces cheap accepted calls, and the difference matters.
Google’s own case work is real and limited. ACKO, an Indian insurer, reported “a 15% increase in cross-channel leads and a 26% reduction in account-level cost per acquisition” (Google). Read the setup before the numbers: Demand Gen was layered on top of existing Search and Performance Max, so both figures are account outcomes after adding a channel, not a controlled test. No holdout. Google-authored, March 2024, and the surfaces included Shorts but were never Shorts alone.
Google also reports that creator partnerships on Shorts inventory “delivered an average 30% increase in conversion lift for Demand Gen campaigns while maintaining CPA efficiency,” citing its own global data from January 2025 to January 2026 (Google). Unaudited internal data, with no comparison arm defined.
Both are directional evidence. Neither is a forecast, and neither is your account.
So this page cannot give you the number you came for. We could not find a defensible published cost per accepted call for Shorts, and we will not invent one. What it gives you instead is the campaign type to open, the constants to cut to, and the measurement that produces your own number in a few weeks of honest spend.
Frequently asked questions
Can I run ads on YouTube Shorts?
Yes. Google documents an explicit Shorts checkbox in Video view campaigns (Google), and Demand Gen serves on Shorts as one of seven ad-group channels (Google). What you cannot assume is that selecting it confines your spend to it.
Can I target YouTube Shorts only?
Google’s documentation answers this two ways and we will not claim one is correct. The channel-controls page gives serving “on YouTube Shorts only” as an example (Google). The Shorts buying page states you “will not be able to buy only YouTube Shorts ads with Demand Gen campaigns” (Google). Set the control, spend, then read your placement report. That is the only thing that tells you what happened in your account.
Does Demand Gen have a call button or a lead form?
No. Google lists both as unsupported: “Click-to-call assets not supported” and “Lead form assets not supported” (Google), and the asset specification contains no phone asset at all (Google). Every call starts on your own landing page and comes back as an offline import.
How long should a Shorts ad be for a call offer?
Work in the 10 to 30 second range Google recommends for action-oriented Shorts (Google). Below 5 seconds you cannot serve, and below 10 seconds you lose in-stream eligibility (Google). A longer cut is fine if the offer needs explaining, but the ask has to land inside the first minute.
What does a Shorts lead or call cost?
There is no defensible public answer. The platform that knows the placement has no call asset, so it never learns which calls were accepted, and the networks that do know acceptance do not publish placement. Judge this channel on your own cost per accepted call, and treat any published Shorts cost-per-thousand figure as unsourced until you can see its method.
If you would rather hand the build to a team that runs this for a living, our YouTube ads management is where to start. We run it for HVAC lead generation and U65 private health lead generation.
