Most content benchmarks are useless because they report an average. Short-form video does not have a meaningful average. We published the real numbers behind ten client videos so you can see the actual shape of it, and the shape is the whole lesson.
The numbers, before the argument
These are the ten videos we publish on our own showcase, with the view counts as reported by the accounts that posted them. Every one was distributed organically. No ad money was spent on any of them.
| Rank | Views | Share of total | Cumulative |
|---|---|---|---|
| 1 | 23,600,000 | 46.0% | 46.0% |
| 2 | 7,488,888 | 14.6% | 60.6% |
| 3 | 6,400,000 | 12.5% | 73.1% |
| 4 | 5,720,000 | 11.2% | 84.3% |
| 5 | 3,330,000 | 6.5% | 90.8% |
| 6 | 2,410,000 | 4.7% | 95.5% |
| 7 | 1,270,000 | 2.5% | 98.0% |
| 8 | 568,000 | 1.1% | 99.1% |
| 9 | 348,000 | 0.7% | 99.8% |
| 10 | 125,100 | 0.2% | 100.0% |
Total: 51,259,988 views. Mean: 5,125,998. Median: 2,870,000.
One honest caveat before anyone quotes this. This is our published showcase, which means it is a selected set of strong work rather than a random sample of everything we have ever shipped. That selection makes the concentration more striking, not less: even after filtering for the good ones, the distribution is still brutally top-heavy.
Share of total views, by video rank
Ten client videos, 51,259,988 views combined. The first bar is one video.
Why the average is the wrong number
The mean of that set is 5.1 million. The median is 2.87 million. When the mean sits at nearly twice the median, the average is being dragged by the tail, and quoting it describes a video that does not exist.
This matters because of how content gets sold and bought. An agency says "our average video does five million views". A client reasonably hears "each video I pay for will do about five million". Then eight of the ten come in under the average and the relationship sours, even though the work did exactly what this kind of work does.
The honest version is that short-form output is a portfolio. You are not buying ten copies of the average. You are buying ten lottery tickets with wildly different payouts, where the whole point of volume is to give the rare enormous one a chance to exist at all.
If your content plan only works when every video performs, you do not have a content plan. You have a wish.
The part that rules out the easy explanations
Here is the finding that actually surprised us.
Four of those ten videos are for the same client, a car dealership in Vaughan, just north of Toronto. Same business, same account, same crew, same broad approach.
| Video | Views |
|---|---|
| Flying salesperson | 23,600,000 |
| Oil money dealership | 6,400,000 |
| Asian dads skit | 5,720,000 |
| No expensive restaurants | 125,100 |
Best to worst is a 189x spread, inside one account.
That number does the heavy lifting in this whole article, because it eliminates almost every comfortable explanation. It was not follower count: identical account. Not the niche: identical niche. Not the production budget, the camera, the editor or the posting time: all effectively the same.
What is left is the idea. The premise of the video, and the first two seconds that sell it.
That is genuinely good news, because the idea is the one input you can change for free. You cannot buy your way to a bigger account overnight, but you can throw away a weak premise before you shoot it.
What this means for how you plan
Stop forecasting per video. Forecast per quarter. The right unit of measurement is "what did ninety days of content produce", not "what did Tuesday's post do".
Publish enough to let the tail happen. If the top outcome is roughly two hundred times the bottom one, a plan with four videos a month is mostly noise. You need enough attempts for the rare one to show up.
Kill weak ideas before the shoot, not after. The expensive part of a bad video is not the edit. It is the shoot day you spent on a premise that was never going to travel.
Judge the system, not the individual result. A video that does 125,100 views is not a failure inside this distribution. It is the cost of the one that does 23.6 million.
How we use this internally
It changes what we argue about with clients. We spend very little time debating whether a finished cut should be four seconds shorter. We spend a lot of time, before anything is filmed, on whether the premise has any reason to travel beyond the people who already follow the account.
We track that with an outlier multiplier rather than raw views: a video's views divided by the median of the account that posted it. Raw views reward big accounts. The multiplier tells you whether the idea did the work. We publish a ranked version of that for real estate in our list of Reel formats ranked by data.
You can see every video in this article, playable, with its real view count on the portfolio.
FAQ
What is a good view count for an Instagram Reel?
There is no single number, and anyone who gives you one is guessing. The useful measure is the outlier multiplier: views divided by the median views of the account that posted it. Anything above roughly 5x means the idea reached beyond the existing audience. A 50,000 view video on an account that normally does 2,000 is a far better result than a 500,000 view video on an account that normally does 400,000.
Why is the average number of views misleading?
Because short-form performance is heavily right-skewed. In the ten video set above, the mean is 5.1 million and the median is 2.87 million, and a single video supplies 46 percent of all views. The average describes the pull of one outlier rather than typical performance.
How many videos should a business post before judging results?
Enough for the distribution to express itself. Given that the gap between best and worst can be close to two hundred times, judging a programme on four or five videos is measuring luck. A quarter of consistent output, somewhere between twenty-five and fifty videos, is the first point where the numbers mean something.
Does a bigger following guarantee more views?
No, and the dealership data is the clearest evidence. Four videos on one account, with a constant follower count, ranged from 125,100 to 23.6 million views. Distribution on these platforms is driven far more by whether a video holds attention in its first seconds than by how many people already follow the account.
Were any of these views paid for?
No. Every number in this article was organic. We have never run ad spend behind a client video, which is also why the distribution is so uneven: paid distribution flattens the curve, and organic distribution does the opposite.