Skip to content
Primores AI wiki

Average Views on TikTok and YouTube: Why the Typical Video Gets About 100

Average Views on TikTok and YouTube: Why the Typical Video Gets About 100

TL;DR: In the only random sample of YouTube, the median video had 35 views, 65.44% had fewer than 100, and the 3.67% of videos with 10,000+ views took 93.61% of all views. On TikTok, a near-complete one-hour slice found a mean of about 2,200–2,500 views per post but a most common value of about 100. The “average” is pulled up by a handful of outliers, so it describes almost no real video.

Because a single post’s views are close to a lottery, Primores’ organic reach programme prices a views target across many TikTok, Reels and Shorts posts rather than per piece, so how many posts it takes to reach the target is our risk, not the buyer’s.

⚠️ Sources read 2026-11-09. Both view-distribution studies are peer-reviewed (Tier 1) and were read in their primary text (PDF / arXiv HTML). TikTok’s and Instagram’s ranking statements are official (Tier 1), read on the platforms’ own pages. No Shorts-only or Reels-only random sample exists, so the YouTube figures cover all YouTube videos, not Shorts specifically.

What “average views” actually means here

Two numbers answer the question, and they disagree by a factor of 100.

MeasureYouTube (McGrady et al. 2023)TikTok (Steel et al. 2026)
Mean views per video5,8682,533 (one-hour set) / 2,244 (one-minute-per-hour set)
Median views35not published
Most common value (mode)not reportedabout 100
Zero views4.88%not reported
Under 100 views65.44%not reported
Under 1,000 views86.93%not reported
Share of views held by top videos3.67% of videos (10k+ views) hold 93.61%not reported
Scalerandom sample of YouTube~269M posts estimated for the day studied

The mean is what people usually quote as “average”, and it is the least useful figure. In the YouTube sample, the 16 most-viewed videos (0.16% of the sample) drew more views (50.52%) than every other video combined. Take those 16 out and the mean collapses. The median (35) and the TikTok mode (about 100) are much closer to what an ordinary post from an ordinary account gets.

Two caveats on the YouTube numbers. The sample covers all YouTube uploads, long videos included, not Shorts alone. And it was drawn before YouTube changed how it counts views: Shorts views have counted every start or replay since 31 March 2025, and all formats count a view at play start since 24 August 2026. Today’s raw counts for the same videos would likely be higher. Our read: the shape (a small typical count, a long tail of outliers) is what transfers, not the exact 35.

Why views are heavy-tailed

The YouTube authors found that views correlate only very weakly with channel subscribers (r = .11), with “many instances of videos with a large number of views uploaded to channels with few subscribers.” Distribution is decided per video, not inherited from the account.

TikTok said the same thing directly in 2020: “neither follower count nor whether the account has had previous high-performing videos are direct factors in the recommendation system.” It also said that finishing a longer video counts as a strong signal and gets more weight than weak ones. A recommender that hands out reach per video, on how viewers respond, will give most videos little and a few videos a great deal.

Instagram is different. Its 2023 ranking explainer lists “information about the person who posted”, including “number of followers or level of engagement”, as an input for Reels, while noting that most of what people see in Reels comes from accounts they don’t follow. So the “a new account can win” premise holds best on TikTok and should not be assumed for Reels.

Topic matters too. In the TikTok slice, posts on medical topics averaged 28.8k views, against a baseline mean of about 2.2k, and posts from bot-like accounts averaged 1.9 views.

What this means if you promise or buy views

One post predicts almost nothing. With a typical count near 100 and r = .11 between subscribers and views, the next video from the same account can do 50 views or 500,000. Our read: a contract or forecast built on “views per post” is built on the wrong unit.

Totals come from outliers. Our read, as arithmetic, not the papers’ claim: at TikTok’s mode of about 100 views, 1M views would take on the order of 10,000 posts. Nobody reaches 1M that way. In practice a batch reaches it because a few posts break out and the rest contribute little.

So pool the risk. Our read: any honest views target is set across a batch of posts and a defined window, with the metric named (raw views vs engaged views) and the read date fixed, because the platforms adjust counts after the fact. A fresh account can still get there: the wiki’s recipes case reached 2.2M views in 30 days with no ad spend (marketing/organic-content-strategy). That is an outcome of volume and testing, not a per-post guarantee. marketing/discovery-before-scale covers the test-then-scale logic.

FAQ

What are the average views on a TikTok video?

A study of a near-complete one-hour slice of TikTok found a mean of about 2,200–2,500 views per post, but the most common value was about 100 views. The mean is inflated by a few very large posts; about 100 is the better guess for a typical one.

What are the average views for a YouTube Short?

No random sample of Shorts alone exists. Across all of YouTube, the median video had 35 views and 65.44% had fewer than 100, measured before the 2025–2026 changes that made every play start count as a view.

Does subscriber or follower count predict views?

Barely, on YouTube: the correlation was r = .11. TikTok says follower count is “not a direct factor”, while Instagram says follower count is one of the inputs for ranking Reels.

Are TikTok views real?

Most are counted from real plays, but views can be bought: researchers found fake YouTube views selling for about €2 per 1,000, and a 2026 vendor check of two sellers found $1.40–$6.70 per 1,000 across TikTok and YouTube. In the US, the FTC’s 2024 rule bans buying or selling views “generated by a bot”.

Can a new account get a viral video?

Yes, it happens, because distribution is decided per video and the YouTube data shows high-view videos on low-subscriber channels. But it is rare by definition: only 3.67% of sampled YouTube videos passed 10,000 views.

Do not cite

  • “TikTok shows every new video to 300–500 (or 200–800) people first.” No official TikTok source names a test-batch size; the figure traces to uncited Douyin explainers copied into English blogs.
  • “The median YouTube video has 41 views.” The paper says 35. The 41 comes from secondary coverage.
  • “YouTube tests each Short on a small seed audience first.” Found only in vidIQ and blog summaries, not in YouTube’s own words.
  • “Follower count doesn’t matter on any short-form platform.” True only of TikTok (“not a direct factor”). Instagram lists follower count as a Reels input.
  • “TikTok posts 117 million videos a day.” Conflicts with the peer-reviewed estimate of 269M posts on the day studied; cite the paper.
  • “X% of short-form views are fake.” No credible prevalence estimate exists; platforms publish removals only.

Key Takeaways

  • The median YouTube video in a random sample had 35 views; 65.44% had under 100 and 4.88% had none.
  • 3.67% of YouTube videos (10k+ views each) held 93.61% of all views; the top 16 videos out-drew everything else combined.
  • On TikTok the mean is about 2,200–2,500 views but the most common count is about 100, across an estimated 269M posts a day.
  • Subscribers barely predict views (r = .11). TikTok says follower count is not a direct factor; Instagram says it is an input for Reels.
  • Our read: price or forecast views across a batch and a window, never per post.

Sources

By Andrej Ruckij