On Twitch, chat doesn't clip: 36% of active channels got zero clips in 30 days.
We measured the community clips of 53,511 Twitch channels streaming in French, English, German and Spanish, seen live between 2026-07-28 and 2026-09-11 with 10 to 4,999 viewers. A community clip is the one a viewer creates with Twitch's Clip button: it is the raw material of short-form content, and the most honest measure of what an audience finds worth sharing. Here is what the numbers say, aggregated by audience size, language and category. No channel is named.
Clips come with the audience, not before it.
Under 50 viewers, 41.9% of channels got no clip in 30 days and the median channel got 1. Between 200 and 999 viewers, the share without clips drops to 24.7% and the median rises to 5. Above 1,000, only 14.4% remain clipless and the median reaches 16. Community clips are a product of size: they don't precede growth, they follow it. A small channel counting on its chat to produce short-form content waits a median 100 hours of live between two clips.
Share of channels with zero clips in 30 days, by observed peak audience
At equal audience, French channels are the least clipped.
All sizes together, 45.9% of French channels got no clip, against 31.1% of English-speaking channels, 41.6% of German and 39.3% of Spanish-speaking ones. The gap is not about size: between 200 and 999 viewers, 31.4% of French channels stay clipless against 21.8% of English-speaking ones, and only 28.8% pass 10 clips, against 38.6%. Between 1,000 and 4,999 viewers, the gap holds: 22.1% without clips on the French side, 11.2% on the English side. The clip reflex is cultural as much as mechanical.
Share of channels with zero clips in 30 days, by channel language
Channels with 200 to 999 viewers: share without clips and share with 10+ clips
A clip above 1,000 views: 1.5% of channels.
Over 30 days, 1.5% of active channels have at least one community clip above 1,000 views. Under 50 viewers it is 0.0%; between 50 and 199, 0.4%; between 200 and 999, 4.0%; above 1,000, 20.9%. A channel's best clip reaches a median 45% of its viewer peak under 50 viewers, and 19% above 1,000: the community clip stays inside the audience that created it and almost never exceeds it. Leaving Twitch, the part that recruits, happens elsewhere, in other formats.
Share of channels with at least one clip above 1,000 views in 30 days
IRL and Just Chatting get clipped. Art, music and WoW never do.
Spoken content is what chat cuts. In IRL, the median channel has 7 clips in 30 days and 13.6% of channels have a clip above 1,000 views, the highest rate of any category. Just Chatting follows with a median of 4 clips. At the other end, 51.6% of Art channels, 49.0% of World of Warcraft channels and 42.6% of Music channels got no clip at all: these are streams where you watch someone work, farm or play music, with no moment to isolate. Competitive games sit in between, with a median of 2 to 4 clips.
| Category | Channels | No clip | Clips, median | Clips, top 10% | Clip > 1,000 views |
|---|---|---|---|---|---|
| IRL | 589 | 22.4% | 7 | 85 | 13.6% |
| Marvel Rivals | 704 | 23.4% | 3 | 28 | 1.1% |
| Dead by Daylight | 1,198 | 24.1% | 4 | 26 | 0.3% |
| Call of Duty: Warzone | 593 | 24.1% | 3 | 21 | 0.8% |
| Halloween: The Game | 647 | 27.2% | 4 | 29 | 0.3% |
| VALORANT | 1,358 | 28.1% | 3 | 30 | 1.5% |
| Just Chatting | 6,134 | 28.9% | 4 | 37 | 3.7% |
| Fortnite | 1,416 | 29.2% | 3 | 27 | 1.8% |
| Project Zomboid | 359 | 29.2% | 3 | 24 | 2.2% |
| Onimusha: Way of the Sword | 388 | 29.4% | 2 | 27 | 2.6% |
| WARDOGS | 646 | 31.1% | 2 | 21 | 2.6% |
| Counter-Strike | 592 | 31.4% | 3 | 27 | 2.2% |
| Overwatch | 848 | 31.5% | 2 | 24 | 0.9% |
| League of Legends | 1,418 | 32.6% | 2 | 24 | 1.6% |
| The Blood of Dawnwalker | 767 | 33.2% | 2 | 26 | 2.2% |
| Minecraft | 1,455 | 33.4% | 2 | 20 | 2.3% |
| Apex Legends | 448 | 33.5% | 2 | 21 | 0.9% |
| FINAL FANTASY XIV ONLINE | 356 | 33.7% | 1 | 12 | 0.6% |
| Special Events | 420 | 33.8% | 2 | 20 | 0.7% |
| Grand Theft Auto V | 1,800 | 35.4% | 2 | 23 | 2.1% |
| Retro | 466 | 36.7% | 1 | 12 | 0.0% |
| Music | 352 | 42.6% | 1 | 11 | 0.0% |
| ASMR | 386 | 43.3% | 1 | 12 | 2.8% |
| World of Warcraft | 1,088 | 49.0% | 1 | 12 | 0.9% |
| Art | 1,041 | 51.6% | 0 | 8 | 0.3% |
Channel's main category (most observed live), categories with 350+ active channels
4.3% of streamers show a TikTok in their Twitch bio.
Discord is the only network truly present in bios (16.5% of active channels). YouTube sits at 5.5%, Instagram and X at 6.7%, TikTok at 4.3%. The smallest channels display TikTok more (6.3% under 50 viewers) than channels with 200 to 999 (1.8%). And the link changes nothing about community clipping: 36.6% of channels with a TikTok in bio got no clip, against 35.8% without. Vertical content is not fed by chat clips.
Out of 777 French channels probed one by one, 628 have a TikTok account and 437 publish on it. Among those whose content could be classified, one third (38%) post stream excerpts; the others publish edited videos or content unrelated to the live. The median posting channel has 37 videos online.
Share of active channels whose Twitch bio links to the network
What changes nothing: chat laughter, and clips on follower count.
Two hypotheses we held and the data rejects. First: a laughing chat would clip more. On 452 French channels whose chat was probed (at least 150 messages), the most laughing quarter (over 9% of messages with a laughter marker) has 21.1% clipless channels and a median of 7 clips; the least laughing quarter (under 4%), 22.2% and 8 clips. No difference. Second: clips would grow the channel. Over the observation window, median follower growth is 0.1% for channels without clips and 0.1% for those with 50 clips or more. The community clip does not recruit on its own; what you do with it elsewhere does.
Where these numbers come from, and what they don't say
Source: Twitch's public Helix API, queried several times a day between 2026-07-28 and 2026-09-11. Population: 72,050 channels seen live with a peak of 10 to 4,999 viewers, channel language declared as French, English, German or Spanish. 71,228 could be enriched between 2026-09-07 and 2026-09-11: followers, published VODs (count, hours, days) and community clips of the last 30 days (count, average views, best clip). The study covers the 53,511 active channels, meaning at least 4 days and 4 hours of VOD (6,492 French, 30,695 English-speaking, 7,715 German, 8,280 Spanish-speaking).
Known limits. The clips counted are community clips created on Twitch, not vertical videos published on TikTok, YouTube or Instagram, which Twitch cannot see. Audience brackets are based on the peak observed during the window, not an average. Language is the one declared by the channel. Follower growth is measured over two to six weeks depending on the channel: it says what happens short term, not over a year. Chat and TikTok probes only cover French channels. The category is the one where the channel was most often seen live. No statistical test is claimed: these are descriptions, at a scale where the gaps read without one.
Data freely reusable (CC BY 4.0) with a link to this page. The full table, by language, audience bracket and category, is available as a CSV. No channel appears in it: the study publishes aggregates, not profiles.
Going further: what separates a good clip from a bad one · the Live Watcher, which clips every stream as it ends.
Questions about the study
What exactly does this study count as a “clip”?
Twitch community clips: the ones a viewer, a moderator or the streamer creates with the Clip button during or after the stream, over the 30 days before the sample. Not vertical videos published on TikTok, Shorts or Reels, which Twitch cannot see.
Why are French channels clipped less?
The data shows the gap, not its cause. At equal size, a French channel is more likely to end the month without a clip than an English-speaking one. Hypotheses consistent with the numbers: an audience less used to the Clip button, fewer clipping moderators, and a younger paid-clipper ecosystem in France.
Can a streamer find their channel in this data?
No. The page and the CSV only publish aggregates by language, audience bracket and category. The source data is public (Twitch API), but we publish no individual profile.
Can I cite these numbers?
Yes, freely, with a link to this page and the sample date. State the scope: Twitch community clips, channels with 10 to 4,999 viewers, four languages, 30 days. The study is updated with each new sample.
Your chat doesn't clip for you. Nysos does, at the end of every stream.
Connect your Twitch or Kick channel: when the stream ends, the Live Watcher reads the replay, grades the moments A→F and renders captioned 9:16 clips. Without waiting for someone to press the button.