From September 16 to October 5, 2026 we published 38 AI-made YouTube Shorts on seven of our own test channels, every one built with Vidpal Studio. By October 5 they had 13,538 public views, a median of 36.5 per Short. One 15-subscriber channel produced 11,026 of those views, while five small channels, mostly brand-new or long dormant, managed a median of 12 views per Short. Raw Studio story Shorts held viewers far longer (60.7% and 69.9% average view duration on the two best) than our tutorial format (27% where measured), and the largest audience group YouTube found was males aged 13 to 17.
This case study is for faceless creators and marketers deciding whether AI-generated Shorts are worth a daily posting habit. The channels were our own test channels, not customer accounts, and they published two kinds of video: raw Vidpal Studio output and Vidpal tutorials showing how those videos were made. We cover the setup, a per-channel scorecard, the format comparison, the audience problem, what we could not measure and what we would change. Every number is the platform's own count, captured on October 5, 2026, and sample sizes are stated with each finding.
Related guides: the 30-day faceless content plan, how to improve retention on faceless videos and the best faceless YouTube niches in 2026.
What did we test, and how?
We tested one question: can AI-generated Shorts, posted on a daily schedule, find an audience without a face, an existing following or ad spend? From September 16 to October 4 we uploaded Shorts to seven YouTube channels we control, usually at 5 PM, and read every number from YouTube on October 5. Another 32 Shorts were scheduled through October 14; they had no data yet and are left out of everything below.
Two formats went out. A raw Studio Short is a finished faceless video exactly as Vidpal Studio rendered it: a narrated story or explainer with AI-generated images and animated captions, uploaded as-is, 56 to 100 seconds long in this test. A tutorial wrapper is a 38 to 53 second Short that opens on a popular channel's page in a niche, shows a one-sentence prompt going into Studio and ends on the finished video. From September 28 on, tutorial titles followed one template: “[Niche] channels get [N]M views. Here's one from a single sentence.”
Three YouTube metrics matter in this post. Stayed to watch is, in YouTube's words, “the percentage of times viewers stayed to watch past the initial seconds of a Short” (YouTube Help). Average view duration is the average time watched among those who stayed to watch, calculated from engaged views (YouTube Help on audience retention). A public view, the number under every Short, is counted the moment a video starts to play, so it is the loosest of the three.
The scorecard: 38 Shorts on seven channels
Here is every channel, described rather than named, with the campaign Shorts it published and the public views those Shorts had on October 5. Subscriber counts are from the same day, and watch hours come from YouTube Studio's 28-day window (September 7 to October 4).
| Channel | Subscribers | What it posted | Shorts live | Views | Median per Short | 28-day watch hours |
|---|---|---|---|---|---|---|
| A: small tutorial channel | 15 | Tutorials | 16 | 11,026 | 689 | 20.8 |
| B: dormant channel relaunched with stories | 4 | Raw Studio stories | 3 | 1,220 | 10 | 7.1 |
| C: established tech channel | 177 | Tutorials, a clip promo, an AI-tools series | 4 | 58 | 8.5 | 6.7 (mostly 2024 videos) |
| D: dormant channel relaunched with stories | 3 | Raw Studio stories and explainers | 6 | 551 | 12.5 | 1.5 |
| E: brand-new tutorial channel | 2 | Tutorials | 2 | 489 | 244.5 | 0.8 |
| F: young brand channel | 2 | Mostly tutorials | 5 | 188 | 20 | Not captured |
| G: young brand channel | 24 | Tutorials | 2 | 6 | 3 | Not captured |
| All seven | — | — | 38 | 13,538 | 36.5 | — |
Two notes on the totals. Our internal review rounded YouTube campaign views to about 14,600; adding up the 38 Shorts one by one from the public counts gives 13,538, which is the figure we use because every line can be checked. And the five small channels (B, D, E, F and G) had between 2 and 24 subscribers on October 5: two had been dormant since 2022 and were relaunched for the test, one was brand-new, and two had started posting in late summer 2026.
How many views does an AI-generated Short get?
In our test, a typical one got very few: the median across all 38 Shorts was 36.5 views, and 17 of them finished below 20. The average, 356 views, is misleading because the distribution is so lopsided. Ten Shorts passed 600 views and eight passed 1,000, and nine of those ten ran on the same channel.
Views were concentrated, not spread. Channel A's 11,026 views are 81% of the total. Channels B, D, E, F and G together posted 18 Shorts for 2,454 views, a median of 12, and three Shorts produced 2,111 of those. The established channel C, with 177 subscribers, did worst per Short: 58 views across four campaign Shorts. Its 854 views that month came mostly from older 2024 videos on unrelated topics, so its subscribers had little reason to watch AI tutorials.
If you take one planning number from this section, take the median. A new AI Shorts channel should expect dozens of views per Short at first, with the occasional breakout, and should judge itself on retention rather than on the rare hit.
Why did one channel get 81% of the views?
Because YouTube kept widening the audience for its Shorts for two weeks, then stopped. Between September 16 and 30, channel A posted 12 Shorts, and 9 of them passed 600 views: 1,112, 1,521, 1,588, 1,275, 746, 1,310, 632, 1,053 and 1,094. The styles were stick-figure animation, simplified cartoon history, mystery storytelling, tech history, Bible stories, philosophy explainers, line-drawing science, city design and anime theory.
From October 1 to 4 the same channel's next four Shorts got 64, 162, 2 and 6 views. YouTube Studio's typical-performance range for a Short of that age on this channel was 250 to 500 views, so these landed 10 to 100 times under it.
Retention explains most of it. The last Short that broke out, on September 30, had 29.0% stayed-to-watch and reached 1,094 views. The next two had 23.4% and 15.9% and stopped at 64 and 162. Earlier Shorts on the channel had been around 31%. YouTube appears to show each new Short to a small test audience first and widen distribution when those viewers stay; our working read is that somewhere below 25% stayed-to-watch, the widening stops. That threshold is our inference from one channel, not a YouTube figure, and with three data points it is a pattern rather than proof: the 15.9% Short outdrew the 23.4% one.
The watch-time math says the same thing. Channel A logged 20.8 watch hours on 11,100 views in its 28-day window, under 7 seconds per view on Shorts that ran 40 to 53 seconds. Most of the people who saw these tutorials swiped early.
Raw AI stories vs tutorials: which format held viewers?
Raw Studio Shorts held viewers far longer. The finished story, with no wrapper, kept 60.7% and 69.9% average view duration on the two strongest, while the tutorial format kept 27% on the one tutorial where we captured it. The table lists every Short we have retention data for. It is seven videos, so read it as a direction, not a benchmark.
| Short (topic) | Format | Channel subscribers | Length | Views | Average view duration | Stayed to watch |
|---|---|---|---|---|---|---|
| The night hip hop was born, Bronx 1973 | Raw Studio story | 4 | 0:56 | 1,209 | 0:39 (69.9%) | Not captured |
| How Marcus Aurelius stayed calm | Raw Studio story | 3 | 1:10 | 89 | 60.7% | Not captured |
| Why you can't stop procrastinating | Raw Studio explainer | 3 | 1:05 | 419 | 39.0% | Not captured |
| History channels, made from one sentence | Tutorial | 2 | 0:40 | 483 | 0:11 (27%) | 41.4% |
| Anime theory channels, made from one sentence | Tutorial | 15 | 0:42 | 1,094 | Not captured | 29.0% |
| Retro gaming channels, made from one sentence | Tutorial | 15 | 0:41 | 64 | Not captured | 23.4% |
| Art history channels, made from one sentence | Tutorial | 15 | 0:40 | 162 | Not captured | 15.9% |
The channel-level numbers point the same way. Story channel B logged 7.1 watch hours on 1,213 views, about 21 seconds per view, three times channel A's rate. The hip hop Short drew 97.9% of its traffic from the Shorts feed, meaning YouTube chose to show it, and its views went flat after about eight hours.
Why did the tutorials get swiped? The opening. A tutorial began by scrolling a famous channel's page, which means nothing to a viewer who doesn't know that channel, and only then revealed the finished video. A raw Short starts with the story itself. On October 5 we switched the tutorials to a result-first opening: the finished video plays first, the explanation comes after. The snapshot in this post predates that change, so we have no data on it here.
One caution in the other direction: raw Shorts are not automatically sticky. The procrastination explainer held 39.0%, and six of the nine Shorts on the two story channels got 15 views or fewer. Format sets the ceiling; it does not guarantee distribution. For techniques that lift retention, see our guide to improving retention on faceless videos and these scroll-stopping hook examples.
Who watched our AI Shorts?
Teenagers more than anyone else, and that is the finding that hurt most. On channel A, the only channel with steady distribution, YouTube Studio's top demographic, both for the channel and for its anime-theory Short, was males aged 13 to 17, with 99% of viewers in the United States. New viewers made up 99% of the audience and regular viewers 0%. Traffic came 83.9% from the Shorts feed and 13% from search.
For a product built for creators and marketers, those views are not leads. Hooks about anime, gaming and “AI made this” were served to the people most likely to enjoy them for a few seconds and least likely to run a channel. Subscriptions confirm it: channel A gained 10 subscribers in its 28-day window on 11,100 views, under 0.1%.
The lesson applies beyond our test. YouTube matches a Short to whoever keeps watching, not to whoever you had in mind. If the first line doesn't say who the video is for, for example “If you run a faceless channel…”, the feed finds its own audience, and for broad, visual AI content that audience skews young. Naming the viewer in the opening line was the next experiment we queued.
Do brand-new YouTube channels get views on Shorts?
Barely, in our test. On the five small channels (B, D, E, F and G, with 2 to 24 subscribers each), most uploads got what looked like a small test, typically 2 to 8 views, and no second push. The two channels set up under our brand name posted seven Shorts between them and collected 194 views.
The exceptions were real but rare: the hip hop story (1,209 views on a 4-subscriber channel), a history tutorial (483 views on a 2-subscriber channel) and the procrastination explainer (419). Those three account for 2,111 of the 2,454 views the five small channels earned. A new channel can break out with one Short; it just can't plan on it.
Channel A was hardly established either: its first upload was in mid-August 2026, a month before the test, and it had 15 subscribers by October 5. Our best explanation for the gap is that its first campaign Short, posted on September 16, broke out (1,112 views), and the channel kept getting wider tests while retention held. With seven channels we can't separate that from luck.
Search did not rescue new uploads either. The first part of a new AI-tools series on the established channel got 8 views, 87.5% of them from search and 12.5% from the Shorts feed, which means the feed never picked it up. If you are starting from zero, plan for weeks of low numbers; our guide to growing a channel from zero covers what to do in that stretch.
Instagram and TikTok told a similar story. We cross-posted the same videos as Reels on five Instagram accounts, and 29 Reels drew about 4,200 views. An aged account with 36 followers got 150 to 200 views on most Reels, and one Reel reached 2,163. The two accounts created for the campaign, with 0 followers, got 166 and 29 views across four Reels each.
TikTok was the weakest platform: 12 live videos on six accounts earned about 260 plays, and four of the six accounts had 0 to 1 followers and 0 to 14 plays. The pattern matches YouTube. Platforms give new, inactive accounts almost no reach, and aged accounts with even a few dozen followers start well ahead.
Did posting on seven channels multiply reach?
No, it divided it. Eighteen Shorts spread across five small channels earned 2,454 views; the 16 on channel A earned 11,026. The three channels we had opened most recently for marketing produced 683 views across nine Shorts, and 483 of those came from a single Short.
There is also a risk we underrated. Seven channels run from one machine, uploading similar templates at the same hour each day, can look like a coordinated network. YouTube's spam policy applies explicitly to “coordinated networks of channels” and lists automated or synthetic mass-production, using AI “to churn out high volumes of similar content with minimal changes”, as a violation. We saw no enforcement during the test, but we moved to one distinct format per channel and fewer, more different videos.
What could we not measure?
The number that matters most: signups. Not one link in more than 100 campaign posts carried a UTM tag, so we cannot say whether any of these views became Vidpal users. UTM tags are the utm_source, utm_medium and utm_campaign parameters that analytics tools read from a link, and Google's documentation says you “should always use utm_source, utm_medium, and utm_campaign” when you tag one (Google Analytics Help). Skipping them turned a marketing test into a reach test.
Other limits to weigh before you reuse these numbers:
- Retention data covers 7 of the 38 Shorts.
- Every figure is one snapshot from October 5, so Shorts posted on October 3 and 4 had only a day or two to grow.
- Public view counts include every start, and average view duration counts only viewers who stayed past the first seconds.
- Seven channels is not a controlled experiment: format, channel age, topic and audience all varied at once.
- The tutorials promoted our own product, which is a harder sell than a neutral story.
What would we do differently?
Almost everything about distribution, and very little about production. The Studio output itself held viewers; the way we packaged and spread it did not. This is the plan we wrote on October 5 for the next round:
- One primary channel per platform, posted to every day, with the others left dormant instead of splitting uploads across five accounts.
- A UTM link in every description, bio and pinned comment, one utm_campaign per account, with signups by source as the weekly number.
- Judge each Short by stayed-to-watch, with a 30% target, and drop any format that misses twice.
- Open on the finished video; proof such as a famous channel or a view count comes second, if at all.
- Name the viewer in the first line, for example “If you run a faceless channel, this takes two minutes.”
- Put raw story Shorts on their own channel, titled as stories, and mention the tool only in the description.
- Never upload near-identical templates to several channels at the same hour.
- Start on aged accounts where possible, and give new ones weeks before judging them.
None of this needs a bigger budget. The test produced 38 Shorts in under three weeks; the plan for the following month cut output roughly in half and aimed it at a third of the accounts.
How to run your own three-week Shorts test
Use the same structure without our mistakes. Each step below takes our lessons as the default.
Step 1: Pick one niche and one channel. Choose a niche where a one-minute story can stand on its own, using our ranking of faceless YouTube niches if you are undecided, and post to a single channel.
Step 2: Make 15 to 21 Shorts before day one. Batch the scripts and renders so the daily slot never slips; the 30-day faceless content plan has a calendar you can copy.
Step 3: Tag every link. Give the channel its own utm_campaign value and use it in the description and the pinned comment of every Short.
Step 4: Post at one time each day. We posted at 5 PM; our guide to how often to post Shorts, Reels and TikToks covers cadence by platform.
Step 5: Read retention before views. YouTube says retention data typically takes one to two days to process, so check each Short's stayed-to-watch and average view duration after 48 hours. Under 30% stayed-to-watch, rewrite the opening of the next one.
Step 6: Check the audience after week one. If the top age group or country is not who you want, rewrite the first line of the next Shorts to name the viewer.
Step 7: Review on day 21. Keep the formats that held 30% or more, drop the rest, and compare signups by UTM rather than total views.
Is posting AI Shorts every day worth it?
For learning, yes; for income on a three-week timeline, no. Three weeks gave us clear answers about format, audience and account strategy that planning alone would not have. But 13,538 views is a long way from monetization: the YouTube Partner Program's Shorts path asks for 1,000 subscribers and 10 million qualified Shorts views in the last 90 days (YouTube Partner Program overview).
Daily posting pays off only when each Short holds viewers and someone reads the numbers every week. That is the honest answer to the autopilot question too: our companion piece, can AI run a YouTube channel on autopilot, covers which parts you can hand to software and which you can't. If you want the failure modes in one place, read why faceless channels fail.
How we made these Shorts in Vidpal
Every video in the test started in Vidpal Studio. For the raw Shorts we typed a topic or a one-sentence prompt, picked a duration and a voice, chose AI-generated images in a style such as stickman, anime, line drawing, 3D cartoon or claymation, added a caption preset and generated. A 9:16 Studio Short costs 1 credit, reserved when you click Generate; the first export is free, and a re-render after changes costs again.
The cost is easy to estimate. Thirty-eight Shorts is 38 credits, just under the 40 credits a month included with Pro ($39 a month, or $23 a month billed annually), and Starter's 15 credits ($19 a month) covers a Short every other day. Studio exports need a paid plan, and publishing from Vidpal, including the Social Scheduler that posts to YouTube, Instagram, TikTok and LinkedIn, is on Pro and Business. Current plan details are on the pricing page.
What Vidpal will not do for you is the part this case study is really about: choosing the niche, writing an opening that names the viewer and reading retention every week. The tool made each video in minutes; the packaging decisions were ours, and so were the mistakes.
Start free on Vidpal to try the workflow. The free plan's 4 one-time credits cover Auto Reels and Image Posts with a watermark, and Studio exports start on Starter.
Frequently Asked Questions
How many views do AI-generated YouTube Shorts get? In our 2026 test, 38 AI-made Shorts on seven channels earned 13,538 public views in about three weeks, a median of 36.5 per Short. Ten passed 600 views and eight passed 1,000, but nine of those ten were on one channel. Five other small channels, with 2 to 24 subscribers each, had a median of 12 views per Short.
Do AI-generated Shorts get suppressed on YouTube? We saw no sign of a blanket penalty. AI-made Shorts passed 1,000 views eight times, including a raw story Short on a 4-subscriber channel. What limited reach in our data was retention and channel history, not the use of AI. YouTube's policies target mass-produced, repetitive content rather than AI itself.
What retention should a YouTube Short have? YouTube publishes no official target. In our data, the tutorial Shorts that broke out had 29% or more stayed-to-watch, and the ones that stalled had 15.9% to 23.4%. Raw story Shorts held 60.7% and 69.9% average view duration. We now aim for at least 30% stayed-to-watch before judging views.
Is it better to post Shorts on many channels or one? One. Five of our small channels posted 18 Shorts for 2,454 views combined, while 16 Shorts on one channel with momentum earned 11,026. Spreading uploads divided reach, and similar templates posted across many channels at once can resemble the coordinated mass-production YouTube's spam policy describes.
Why did YouTube show our AI Shorts to teenagers? YouTube matches Shorts to the viewers who keep watching, and broad visual hooks about anime, gaming and AI held a young audience best. The top demographic on our strongest channel was males aged 13 to 17, with 99% of viewers in the United States. Naming the intended viewer in the first line is the fix we queued next.
How much did the 38 Shorts cost to make? On Vidpal's current pricing a 9:16 Studio Short is 1 credit, so 38 Shorts is 38 credits, within the 40 credits a month on the Pro plan at $39 a month. The tutorial versions also needed screen recording and editing time outside Studio.
Did the AI Shorts bring in any signups? We don't know, and that is the biggest gap in this test. None of the links in the campaign carried UTM parameters, so signups could not be traced to individual channels or videos. Any repeat of this test should tag every description, bio and pinned-comment link before the first upload.
