Kick clip yield: production data from 51 real sessions
Most claims about stream clipping are marketing. This page is measurement. Between June 19 and July 14, 2026, ClipMe's production pipeline processed 51 real capture sessions across 8 Kick channels. Every number below is computed directly from the finished clip files on disk: counts, durations and dates, nothing else. The per-session table is downloadable as CSV and licensed CC BY 4.0, so you can cite it, chart it or check our math.
The headline numbers
613
finished clips delivered
51
capture sessions
30s
median clip length
28%
of clips cut while live
- 613 finished clips from 51 sessions across 8 Kick channels, 2026-06-19 to 2026-07-14.
- 1282 rendered outputs total, since most clips ship in more than one aspect ratio (9:16, 1:1, 4:5, 16:9).
- 173 clips (28%) were cut during the live broadcast, across 14 live capture sessions. The rest came from finished VODs.
- 5.16 hours of finished short-form output; median clip 30s, mean 30.3s.
Clip-length distribution
88% of delivered clips run 30 to 60 seconds. That concentration is deliberate: it is the band where a moment fits a setup and payoff and still holds on TikTok, Shorts and Reels.
| Length | Clips | Share |
|---|---|---|
| Under 15s | 14 | 2% |
| 15–30s | 52 | 8% |
| 30–60s | 540 | 88% |
| 60–120s | 7 | 1% |
| 120–300s | 0 | 0% |
Session-yield spread
Yield varies widely by session: the median session in this dataset delivered 10 clips, with a spread from 1 to 50. The top of the distribution is full-length VOD runs — the three largest sessions delivered 46, 50 and 50 clips — which is where ClipMe's ~10-hour-stream-to-~50-ranked-clips benchmark sits. Live capture sessions average about 12 clips each (173 clips across 14 sessions), and they skew shorter than full VOD runs. The dataset average of ~12 clips per session reflects that mix of short live captures and full-length VODs, not a per-stream ceiling.
Honest limit: v1 of the CSV does not include source-stream durations, so yield-per-hour can't be computed from the published table yet. A duration column is planned for the next dataset release.
Two reference runs
- The 24-hour run. A production run outside this dataset's window (not among the 51 sessions in the CSV): one 24-hour Kick VOD produced 69 delivered clips.
- The speed benchmark. End-to-end pipeline throughput, measured end-to-end on production hardware: a ~10-hour stream ingested, scored, ranked and cut into ~50 finished clips in about 5 minutes of wall-clock time — roughly 120× real-time (real-world varies with stream length, queue and plan).
On the human side of the pipeline: a typical VOD in this period surfaced roughly 25 picks worth reviewing, and the review-and-post loop ran about 6 minutes per stream.
Who builds ClipMe
The pipeline behind these numbers is built by 723 Studios, the engineering studio of ClipMe CTO Steven Malagon — the developer who built The Moron Test as DistinctDev's CTO, one of Apple's Top 20 best-selling iPhone apps of all time, with more than two years in the App Store's Top 100. Founder Samuel Segers owns product, clip-scoring direction and distribution; 723 Studios ships the engineering.
Methodology and limitations
- Every published figure is derived from the shipped media files themselves — each clip measured directly with ffprobe — not from logs, sampling or estimates. Multi-format renders of the same moment are deduplicated, so "clips" counts unique moments, not files.
- Files over 5 minutes (compilation reels and segments) are excluded from clip statistics.
- Channels are anonymized (Channel A through H). Capture sessions include both live HLS captures and VOD processing runs; the CSV marks which is which.
- Honest limits: this is one pipeline's output, not a survey of the category. Clip counts reflect what was delivered after review, not every candidate the engine surfaced. No engagement or views data is included, only production yield.
Download the per-session data: kick-clip-yield-2026.csv (CC BY 4.0, attribution: "ClipMe Research"). If you use it in a story or a chart, a link to this page is appreciated but not required.
Cite this dataset
ClipMe Research (2026). Kick Clip Yield Dataset: 613 clips from 51 live-stream capture sessions, June–July 2026. CC BY 4.0. https://clipme.com/research
FAQ
How many clips does a Kick stream produce?
In ClipMe's production dataset of 51 processed capture sessions (June 19 to July 14, 2026), the pipeline delivered 613 finished clips — a median of 10 per session, ranging from 1 up to 50 for full-length VOD runs. Yield varies widely with session length and how eventful the stream is; a separate 24-hour production run outside this dataset window produced 69 delivered clips.
How long is the typical stream clip?
In this dataset the median delivered clip is 30 seconds, and 88% of all clips land between 30 and 60 seconds. That range fits short-form platform norms: long enough for a setup and payoff, short enough for TikTok, Shorts and Reels retention.
Can clips really be cut while the stream is still live?
Yes. 173 of the 613 clips in this dataset (28%) were cut during live Kick broadcasts across 14 live capture sessions, by reading the live HLS feed rather than waiting for the VOD. The rest were processed from finished VODs.
Where does this data come from?
From ClipMe's own production pipeline: 51 real capture sessions across 8 Kick channels, with every clip file on disk measured directly with ffprobe. Channels are anonymized. The per-session table is downloadable as CSV under CC BY 4.0.
Run these numbers against your own week with the free clip calculator, which derives every rate it uses from this dataset.
Related reading: how Kick clips actually work, the state of AI clipping 2026 and the best AI clipping tools for Kick.
Related
More from ClipMe