Short-Form Video Statistics 2026

If you are here because someone asked you to justify short-form in a budget meeting with a marketing ROI figure, the two short-form video ROI statistics you want are 91% and 49%. 91% of businesses use video as a marketing tool in 2026, an all-time high, and short-form video is the single highest-ROI content format, ranked top-three by 49% of marketers. The first comes from Wyzowl's video marketing statistics for 2026 (its State of Video Marketing report), the second from HubSpot's 2026 State of Marketing. Those two sentences will survive the meeting.

What follows is the part that usually does not make the slide.

The number I find most interesting is the one that went down. Reported ROI slipped from 93% to 82% in a single year, while adoption hit an all-time high. Those two facts only look contradictory if you assume every business using video is using it the same way.

Here is what I think happened, and I will say plainly that this is my read rather than something Wyzowl measured. When 91% of businesses have adopted a format, the population reporting on it stops being early adopters who invested properly and starts including everyone who posted four videos to a company page and called it a channel.

The format did not get worse. The average got diluted. "Highest-ROI format" is a verdict on what short-form can do, not a promise about what any particular team will get out of it.

That distinction matters for you, because it means the competitive edge has moved. In 2020 the edge was using short-form at all. In 2026 almost everyone does, so the edge is how far you can push it past the people who also do. That push is mostly distribution, which is the part a short-form video agency takes on.

TikTok averaged a 3.70% engagement rate in 2025, against 0.48% on Instagram and 0.15% on Facebook. Socialinsider measured that across roughly 70 million posts from international brand accounts, which makes it one of the better-powered benchmarks available publicly. TikTok's rate is more than seven times Instagram's and close to twenty-five times Facebook's.

Now the part that took me a while to accept, and it is the reason this page exists in the form it does. A 3.70% engagement rate is a percentage of the people who already saw the post, and on a single brand account that is mostly your existing followers. It is a quality signal. It is not a reach signal.

Put it this way. Two brands can both hit a 3.70% engagement rate. One has 2,000 followers and one is posting the same clip across two hundred accounts. Their engagement rates are identical and their outcomes are not remotely comparable. The rate tells you the clip landed with whoever saw it. It says nothing about how many people that was.

So I would use the platform gap to decide where a clip goes first, and I would not use it to decide anything else. Cut natively for each feed while you are at it. Horizontal reposts and visible watermarks from another app get quietly suppressed, and that is a self-inflicted wound I see constantly. If you are deciding where to start, the platform breakdown goes deeper on what each feed rewards, and how to clip on TikTok covers the native cut itself. For where a clip gets reach rather than engagement, see Instagram Reels vs TikTok.

TikTok's adult advertising reach passed 1.99 billion people aged 18 and over, and the average user spends around 95 minutes a day in the app, the highest of any major platform DataReportal measures. Worth being precise about what that first figure is: it is the audience the platform reports as reachable with advertising, not a monthly-active-user count. The two get used interchangeably in stat roundups and they are not the same thing.

The practical conclusion from an audience that size is short, and it is the opposite of what most stats pages conclude. For almost any niche, demand is not your constraint. There are enough people on these platforms who would watch your content that you will never exhaust them. Nobody runs out of audience.

Which means when an account stalls, the cause is almost never that the audience is too small or too tired. It is that not enough of your content reached them. That is a supply and distribution failure dressed up as a demand problem, and it is why "make better videos" so rarely fixes a flat account. The videos were often fine.

Reach is roughly the number of clips, multiplied by the median views per clip, multiplied by the number of accounts posting them. Every public benchmark in this article measures the middle term and ignores the other two. That is not a criticism of the research. Wyzowl and Socialinsider are measuring what they can see. But it does mean a brand can read every short-form report published this year and still miss the lever with the most slack in it.

Two of those three terms are distribution and volume. Only one is the thing most teams spend their money on.

A minority of clips drive most of the views. That is true on our campaigns and it is true on yours, and it is the single most important structural fact about this format. You cannot reliably manufacture the breakout. Nobody can, and anybody selling you a method that can is selling you something else.

What you can do is take more shots, from more accounts, with a filter on the way in. That is not a clever strategy. It is just the only one that survives the fact that outcomes are wildly uneven. The mechanics of running it that way are in short-form video distribution, and if the model itself is new to you, what a clipping agency does is the shorter starting point. For where the raw material comes from, turning one long video into clips covers the front half of the pipeline.

Everything above this line is somebody else's research. This section is ours, measured through native platform analytics across live campaigns, and I am going to give you the uncomfortable number as well as the flattering ones.

Across the network we have driven more than 18 billion verified views, with 62,900+ vetted clippers on the roster. Forbes put the figure at 62,000+ vetted clippers plus 5,000 UGC creators in its piece on 23 July 2026, which was a third-party snapshot taken in July. Both can be true: one is our own current count, the other is where a reporter found it in the summer.

Now the number I would rather not publish, and the reason I do. On one campaign, 9,682 clips were submitted and 4,255 were approved. That is a pass-rate of 44%, which means more than half of everything sent to us earned nothing.

I have watched that number make prospects uncomfortable on calls, and I keep publishing it anyway, because a provider who approves everything has no quality bar at all. When you are evaluating anyone in this space, the rejection rate is a more revealing question than the view count. View counts can be bought. A rejection rate is an admission that somebody is checking.

There is a practical reason for the bar beyond pride in the work. Volume without a filter is spam, and platforms are extremely good at recognising spam. Push enough weak clips from enough accounts and you do not get a bigger number, you get suppressed accounts and a smaller one. The floor protects the ceiling.

We have run clipping campaigns for crypto exchanges, iGaming operators, consumer software and global brands, across TikTok, Reels, Shorts and X. The names below are clients whose campaigns we have run, not logos borrowed from a partner list. The fuller write-ups live in our case studies, and the mechanics of how a programme is set up are in clipping campaigns.

You will notice there is no view count next to any of those names, and that is deliberate. We publish a per-client figure only once it has been read off that campaign's own dashboard, with the window it covers attached.

A view total means nothing without knowing whether it is thirty days, one campaign or all time, and the same brand quoted over three different windows produces three numbers that all look like contradictions. Until a figure is confirmed against the dashboard, the brand appears here with no number at all.

What I can show you in full is the shape of a single campaign, because the review queue records every clip that came in and every decision made on it. This is what a brief going out to a clipping network actually produces.

Engagement rate should not be your campaign KPI. It is a good diagnostic and a bad target. Total qualified reach and the actions downstream of it are what you are actually buying. Here is the rest of how I read a short-form report.

In categories where paid advertising is restricted, organic distribution at scale is not a preference, it is the only channel that scales. Crypto and iGaming are the clearest cases. If you cannot buy the reach, the only route left is earning it across enough accounts to matter, which is why those verticals were the first to take distributed clipping seriously.

B2B is the quieter case and it surprises people. Buyers research on the feed for months before they ever speak to sales, and founder-led clips travel considerably further than brand-account posts do. A face outperforms a logo on every one of these platforms. If you are weighing this against paying creators directly, clipping versus influencer marketing versus UGC sets the three side by side on cost and reach.

There is also a second-order effect that did not exist three years ago. The same clip distributed across many accounts seeds brand mentions across many sources, and mentions across many sources are what AI assistants and search summaries are reading now. Wide distribution builds entity signals faster than one viral post does, because breadth of mention outweighs a single spike. If you care how an assistant describes your brand, that is a distribution problem too. Lumina Clippers is a clipping agency that distributes brand videos through 62,900+ clippers.

Every statistic on this page is listed below with the company that published it and what it measures. Nothing here is quoted from a stat-roundup blog. Where a figure is a forecast rather than a measurement, the table says so.

Every figure in that table is attributed to the company that published it and cross-checked across independent reporting of that publication, and every first-party number is measured in our own systems rather than estimated. Nothing on this page is an industry guess, a rounded-up number or a figure borrowed from another site's roundup. Where we could not stand a number up against its source, it is not on the page at all.

The whole table is available as a downloadable CSV, with each row marked first-party or third-party. It is free to cite with attribution and a link back. If you are quoting this page: Lumina Clippers, "Short-Form Video Statistics 2026", luminaclippers.com/blog/short-form-video-statistics. Last reviewed 21 September 2026 by Rhys McKay, founder of Lumina Clippers.