Short-form editing benchmarks by format: cuts, length and loudness from 138 reels
Cuts per minute, length, loudness and reach for all 23 formats we measured, from 138 torn-down reels on 69 pages. Each row says how many pages stand behind it, because formats measured on few pages are hypotheses.
The result
Among formats with Medium or Higher confidence, median hard cuts per minute ran from 0.0 for Code physics simulation and Narrated facts / GK to 18.6 for Hands-only real-process ASMR. On reach, the highest median plays per follower among those formats was 17.38 for Code physics simulation and the lowest was 0.05 for Wisdom, stoic, shayari and quote pages. Loudness barely varies: the median across the 126 reels with audio was -14.2 LUFS, and 92.9% sat within 1 LU of -14 LUFS. The first hard cut came after 3 s in 48.6% of reels, and a further 27.5% had none. Reels with a series number in the caption reached 3x their page median in 24.2% of cases against 21.5% without one, a small gap.
The numbers by format
Rows are sorted by median plays per follower. The length, cut and loudness columns are medians over the torn-down reels of that format; the other columns come from the Format scoreboard.
| Format | Pages measured | Reels measured | Median length (s) | Hard cuts/min | Big visual changes/min | Median loudness (LUFS) | Plays per follower (median) | Reels at 3x+ own median | Confidence |
|---|---|---|---|---|---|---|---|---|---|
| Code physics simulation (race, battle, merge) | 3 | 30 | 49.2 | 0.0 | 0.0 | -14.3 | 17.38 | 7 of 30 | Medium |
| Satisfying escalation ('10 vs 1,000 cuts') | 1 | 10 | 72.0 | 0.0 | 1.2 | -14.4 | 7.56 | 5 of 10 | Low (hypothesis) |
| Interactive survival series (audience votes next episode) | 5 | 47 | 87.1 | 5.7 | 10.6 | -14.1 | 5.45 | 9 of 47 | Higher |
| AI giant-object room reveal (ASMR) | 3 | 28 | 57.4 | 5.8 | 8.1 | -14.4 | 4 | 6 of 28 | Medium |
| Stickman / simple 2D comedy and explainers | 5 | 50 | 23.2 | 14.4 | 10.2 | -14.1 | 2.94 | 19 of 50 | Higher |
| Animal biology explainer | 1 | 10 | 76.2 | 8.4 | 19.3 | -14.2 | 2.63 | 4 of 10 | Low (hypothesis) |
| News stories retold as a cartoon | 1 | 10 | 18.2 | 6.7 | 18.6 | -14.1 | 2.45 | 4 of 10 | Low (hypothesis) |
| Quiz / would-you-rather / pick-one | 2 | 20 | 80.1 | 3.5 | 7.3 | -14.8 | 2.27 | 7 of 20 | Low (hypothesis) |
| Hands-only real-process ASMR (food, craft, care) | 3 | 30 | 29.1 | 18.6 | 13.6 | -15.4 | 2.06 | 11 of 30 | Medium |
| AI ASMR cutting / creepy-cute | 2 | 20 | 8.2 | 3.2 | 0.0 | -22.4 | 1.52 | 3 of 20 | Low (hypothesis) |
| AI cinematic short-film series | 6 | 57 | 26.6 | 13.8 | 16.8 | -14.1 | 1.29 | 19 of 57 | Higher |
| Space and mystery narrated explainer | 3 | 30 | 61.1 | 4.7 | 9.1 | -14.1 | 0.57 | 4 of 30 | Medium |
| Narrated horror story over AI visuals (Hindi) | 3 | 30 | 66.6 | 9.9 | 14.1 | -14.1 | 0.55 | 3 of 30 | Medium |
| AI talking-baby comedy (Hindi and regional) | 6 | 60 | 15.7 | 15.3 | 3.0 | -14.1 | 0.35 | 14 of 60 | Higher |
| History POV / reconstruction | 8 | 70 | 39.8 | 1.1 | 6.8 | -14.1 | 0.3 | 11 of 70 | Higher |
| AI anthropomorphic object / vegetable serial (Hindi comedy) | 3 | 30 | 94.4 | 10.8 | 12.6 | -14.2 | 0.29 | 6 of 30 | Medium |
| AI recurring character IP (pets, oddballs, character universes) | 8 | 60 | 15.3 | 5.6 | 6.8 | -14.2 | 0.28 | 12 of 60 | Higher |
| Finance infographic | 1 | 10 | Not measured | Not measured | Not measured | Not measured | 0.24 | 1 of 10 | Low (hypothesis) |
| AI family melodrama / kahani serial (Hindi) | 4 | 38 | 64.7 | 5.3 | 12.4 | -14.1 | 0.2 | 9 of 38 | Medium |
| Narrated facts / GK (Hindi) | 8 | 76 | 53.2 | 0.0 | 3.7 | -14.2 | 0.15 | 15 of 76 | Higher |
| Mythology and devotional animation | 6 | 52 | 58.8 | 5.7 | 6.7 | -14.2 | 0.1 | 7 of 52 | Higher |
| Edit culture (cars, cricket, film fandom) | 5 | 50 | 32.4 | 0.7 | 4.7 | -14.1 | 0.09 | 7 of 50 | Higher |
| Wisdom, stoic, shayari and quote pages | 6 | 58 | 44.2 | 1.8 | 2.8 | -14.1 | 0.05 | 12 of 58 | Higher |
What the numbers do not tell you
- Why a reel got its plays. The numbers describe what top and typical reels look like, not what caused their reach.
- Retention, shares and follows. These cannot be seen from outside Instagram.
- Whether a format works for a new page. Plays per follower favours small, newer pages, and our pages were found through hashtags, which favour pages that already had a hit.
- Anything about formats with Low confidence. One or two pages are a hypothesis, not a benchmark.
- Anything about dissolves. The hard-cut count misses them, and the big visual change count mixes cuts with camera travel.
- Anything about a page's other reels. We measured the best and a typical reel of each page, and ten reels from one page are not ten independent tests.
How we measured
- Pick the best reel and a typical reel of the last ten from each of 69 pages, 138 reels in all.
- Measure each file by script: length, hard cuts, big visual changes, loudness and speech pace.
- Take plays, followers and posting rate from the Apify collection, then recompute the format table from the raw files.
- Confidence follows the number of pages measured: Low for one or two, Medium for three or four, Higher for five or more.
For one reel measured this way, see the reel teardowns.
Method and limits
- Plays, likes, comments, followers and posting dates were collected from Instagram by Apify on 5 Nov 2026.
- For each of 69 pages we took the best reel and a typical reel of its last ten, 138 reels in all, and measured the files by script.
- Hard cuts, shot length and the first cut come from a script that finds hard cuts between frames.
- Big visual changes count hard cuts plus large camera travel or scene changes, so they are always at least as many as hard cuts and also catch some fast moves.
- Loudness is the integrated loudness in LUFS of the audio track. Speech pace is words per second from a speech-to-text pass, with a language guess and its probability.
- The frame-by-frame notes were written by a model reading contact sheets of each reel. They are observations, and each teardown page labels them that way.
- Ratio to the page median is plays divided by the median plays of the page's last ten reels, as measured on 5 Nov 2026.
- Format statistics are recomputed here from the raw files and compared with the workbook's Format scoreboard; any difference is recorded under recheck.
- Pattern counts per format are keyword searches over the frame-by-frame notes of that format, so read them as approximate.
- Hard-cut counts miss dissolves, so a reel built on dissolves can show close to zero hard cuts.
- The big visual change count also counts camera travel and scene changes, so it is not the same as the cut count. We show both.
- Some full-HD files had no audio track, so loudness and speech are blank for those reels.
- Plays include replays. Shares and follower conversion cannot be seen from outside.
- Ten reels from one page are not ten independent tests.
- Hashtag discovery favours pages that already had a hit, so breakout pages are over-represented.
- Plays per follower favours small, newer pages; it measures reach health, not size.
- Formats measured on only one to three pages are hypotheses.
- No number here is a promise about what a new reel will do.
Plays and followers as shown on Instagram on 5 Nov 2026. We recomputed every scoreboard field from the raw files and found no differences; the check re-runs the same arithmetic on the same files.
Page checked on Nov 4, 2026.