How long app videos run: 30 seconds for a promo, 82 for a listicle
We timed 1,100 videos, 100 in each of 11 formats. The median promo runs 30 seconds and the median listicle 82, and the middle 80% of promos runs anywhere from 7 seconds to 99.
Videos published 20 May to 13 August 2026
- 1,100
- Videos timed
- 30s
- Median promo
- 82s
- Median listicle
1,100 videos, 100 in each of 11 formats · Videos published 20 May to 13 August 2026, snapshot of 14 August 2026
Running time is one of the few things in this dataset we observe instead of estimate. The platform reports the length of a video in whole seconds, and a second means the same thing on TikTok and on Instagram.
We took the 100 most recent videos in each of the 11 content types our classifier uses, published after 16 May 2026. All 1,100 carry a duration. The medians run from 15 seconds to 82, and nine of the 11 sit at 45 seconds or above. 498 of the 1,100 videos ran longer than a minute.
The measurement needs no hedge. The grouping does. Every video is classified by a language model into a fixed vocabulary of formats, hooks, tones and CTAs. It is a classification, not a human tag, and it is spot-checked, not benchmarked — so we publish it as a distribution across hundreds of thousands of posts, never as a judgement about any single video. The seconds are observed; which pile a video lands in is a model output.
Eleven formats ranked by how long they run
Median running time in the 100 most recent videos of each content type, published after 16 May 2026.
Median seconds per format. Each bar is a separate draw of 100 videos, so the bars sum to nothing. No bar is weighted by how common the format is.
The figures behind this chart
| other (residual bin) | 15 |
| promo | 30 |
| storytime | 45 |
| pov | 54.5 |
| testimonial | 55 |
| tips | 56.5 |
| reaction | 60 |
| demo | 66.5 |
| comparison | 67 |
| review | 76 |
| listicle | 82 |
1,100 videos, 100 in each of 11 formats · Videos published 20 May to 13 August 2026, snapshot of 14 August 2026
The order tracks what each format has to do. A promo asserts one thing and lands in 30 seconds. A listicle counts items out loud and takes 82, close to three times as long.
The shortest entry is not a format. other is the classifier's residual bin, and 48 of its 100 videos ran under 15 seconds. It sits on the chart because dropping it would hide where short unclassifiable clips go.
The two commonest formats are the two shortest real ones. Promo and tips account for 323,829 of the 419,998 videos that carry a content-type label, which is 77.1% of them. Their medians are 30 and 56.5 seconds.
The spread inside a format is wider than the gap between formats
The 10th and 90th percentiles of running time in each 100-video draw, with the count that ran over a minute.
Each row is one draw of 100 videos of that content type, ordered by median running time.
| Format | 10th pct (s) | Median (s) | Mean (s) | 90th pct (s) | Over 60s |
|---|---|---|---|---|---|
| other (residual bin) | 7 | 15 | 28.3 | 68 | 11 |
| promo | 7 | 30 | 43.7 | 99 | 25 |
| storytime | 14 | 45 | 65 | 121 | 38 |
| pov | 15 | 54.5 | 70.6 | 125 | 45 |
| testimonial | 24 | 55 | 67.4 | 137 | 42 |
| tips | 13 | 56.5 | 71.4 | 162 | 46 |
| reaction | 18 | 60 | 76.9 | 146 | 49 |
| demo | 11 | 66.5 | 82 | 176 | 55 |
| comparison | 16 | 67 | 86.8 | 152 | 54 |
| review | 31 | 76 | 98.4 | 180 | 71 |
| listicle | 15 | 82 | 94.7 | 179 | 62 |
The mean sits above the median in all 11 formats, because a small number of long videos pulls it up. Use the median and the two percentiles; the mean is printed so the skew is visible instead of hidden.
1,100 videos, 100 in each of 11 formats · Videos published 20 May to 13 August 2026, snapshot of 14 August 2026
A median with no spread beside it invites a confidence the data does not carry. Demo is the widest format in the draw: 11 seconds at the 10th percentile and 176 at the 90th, a 16-fold range. Promo is close behind at 7 and 99.
Two formats are tight at the short end. The 10th percentile of review is 31 seconds and of testimonial 24. One of the 100 reviews ran under 15 seconds, and 2 of the 100 testimonials did. A 10-second review is something 99 of these 100 creators did not publish.
The long tail is where the formats separate hardest. 28 of the 100 reviews and 32 of the 100 listicles ran over two minutes, against 4 of the 100 promos.
- 498
- Of the 1,100 ran over 60 seconds
- 145
- Ran under 15 seconds
- 603s
- Longest single video, a comparison
1,100 videos, 100 in each of 11 formats · Videos published 20 May to 13 August 2026, snapshot of 14 August 2026
TikTok runs longer than Instagram in 8 of the 11 formats
Median running time split by platform inside each draw, with the number of videos behind each median.
The same 1,100 videos, split by platform. Instagram counts fall as low as 10 in a format, so the right-hand median is thin where the count says so.
| Format | TikTok median (s) | TikTok videos | Instagram median (s) | Instagram videos |
|---|---|---|---|---|
| other (residual bin) | 15 | 79 | 13 | 21 |
| promo | 35 | 58 | 25.5 | 42 |
| storytime | 56.5 | 76 | 34 | 24 |
| pov | 54 | 76 | 56.5 | 24 |
| testimonial | 53.5 | 66 | 56.5 | 34 |
| tips | 62.5 | 86 | 36.5 | 14 |
| reaction | 60 | 77 | 66 | 23 |
| demo | 68 | 81 | 44 | 19 |
| comparison | 69 | 75 | 58 | 25 |
| review | 85.5 | 86 | 48 | 14 |
| listicle | 82 | 90 | 71.5 | 10 |
Instagram medians for tips (14 videos), review (14) and listicle (10) rest on counts too small to carry a comparison on their own. They are printed with their n so the reader can discount them.
1,100 videos, 100 in each of 11 formats · Videos published 20 May to 13 August 2026, snapshot of 14 August 2026
The platform gap is widest in tips, 62.5 seconds on TikTok against 36.5 on Instagram, and in review, 85.5 against 48. Instagram runs longer in three formats: reaction, testimonial and pov.
That matters for reading the first chart, because the platform mix is not the same in every draw. 42 of the 100 promos are Instagram, against 14 of the reviews and 10 of the listicles. Some of the promo-to-review gap is the mix and not the format.
So we checked the ranking inside TikTok alone, where the mix cannot move it. The promo median is 35 seconds, the review median 85.5 and the listicle median 82. The short end and the long end hold. Storytime is the one row that moves far, from third-shortest overall to fifth inside TikTok, because its Instagram half runs at 34 seconds.
Brief the length the format already runs at
What the medians support when you commission a video or cut your own, stated as counts out of 100.
Commissioning a demo. The middle of the TikTok draw is 68 seconds and 55 of the 100 demos ran over a minute. A 15-second demo exists: 14 of the 100 came in under 15 seconds. It is a minority of what people making demos this quarter published.
Cutting a promo yourself. 30 seconds is the middle, and the middle 80% runs from 7 seconds to 99. 25 of the 100 promos ran over a minute, so a long promo is ordinary and not an error by itself.
Paying a creator for a review. 71 of the 100 reviews ran over 60 seconds and 28 ran over two minutes. The 10th percentile is 31 seconds, so short reviews are close to absent. Scope the edit and the fee against a minute and a quarter, not against 30 seconds.
Put two numbers in the brief, not one
Give the creator the format's median and its 90th percentile. For a demo that is 66.5 seconds and 176. A single target invites a cut that fights the format. The second number tells the creator how much room they have before they leave the ordinary range.
These times say nothing about which length performs
Four things the measurement cannot see: performance, the label, the draw, and the library behind it.
No length on this page is a recommendation. We counted what exists. We did not measure what worked, and we deliberately print no views or virality beside these medians. The draw would make that comparison misleading anyway: it takes the 100 newest videos of each format, and how common the format is decides how far back that reaches. The 100 newest promos were published in the 10 days to 13 August. The 100 newest comparisons reach back to 20 May. Ranking those two draws by views would compare 10 days of accumulation against 12 weeks.
The grouping is a model output. The seconds are read from the platform. Which of the 11 piles a video lands in is decided by a classifier over its caption and transcript. A video a human would call a demo may sit under tips.
Each format contributes exactly 100 rows, so this is not the library's mix. The library holds 449 comparison videos, so the 100 we timed are 22.3% of every comparison we hold. It holds 212,523 promos, so the 100 we timed are 0.05% of them. Each row describes its own format. Adding the rows together would describe nothing.
The library is a sample. These figures describe Apptonic's own library as of 14 August 2026. The library is a growing sample of the app market, not a census. The draw is library-wide, not filtered to a category, and it includes posts we have not matched to any app.
Nothing on this page is an estimate
Duration is read from the platform payload in whole seconds. It is not modelled, not rounded at source and not derived from any third-party provider, and it carries the same meaning on both platforms. The counts of videos over 60 seconds and under 15 are counted from the same 1,100 rows.
How we counted
Every running time comes from one query per content type, frozen into the wave snapshot of 14 August 2026 as global/duration-by-content-type.json. For each of the 11 values in the classifier's content-type vocabulary we drew the 100 most recent videos published after 16 May 2026, sorted newest first. The draw is library-wide, with no category filter and no target-market filter. The rows drawn were published between 20 May and 13 August 2026. All 1,100 carry a duration, so no row is dropped from a median.
durationSecs is read from the platform payload in whole seconds. Medians, means and the 10th and 90th percentiles are the harvest's own fields. Four figures are counted from the 1,100 rows in the same file: videos over 60 seconds, videos under 15 seconds, videos over two minutes, and the longest single video.
The draw is fixed at 100 per format, which is a different share of each format. The library holds 449 comparison videos and 212,523 promos, so the two draws cover 22.3% and 0.05% of their formats. No figure on this page is weighted by how common a format is, and no figure is a library average.
Format shares come from the global facet call in the same snapshot: 419,998 videos carry one of the 11 content-type labels, against 565,898 videos in the library, which is 74.2%. Those two counters are computed differently and are never mixed inside one ratio.
We print no views, no virality score and no performance figure of any kind beside these durations. The draw is recency-truncated by format and cannot support that comparison. Nothing here is modelled: every figure is a row count or a percentile over row counts.
- Videos timed, 100 in each of 11 content types
- 1,100 rows · published 20 May to 13 August 2026
- Videos carrying one of the 11 content-type labels
- 419,998 rows · facet pass, 14 August 2026
- Organic videos in the library, TikTok and Instagram
- 565,898 rows · library counter, 14 August 2026
Frequently asked
- How long should my app video be?
- This page cannot answer that, and any page that answers it from duration data alone is guessing. We counted what creators published, not what worked. What the counts give you is the length the format already runs at: 30 seconds for a promo, 66.5 for a demo, 76 for a review, 82 for a listicle. Brief against those and you are asking for something ordinary instead of something nobody in the sample made.
- Is the running time an estimate?
- No. Duration is read from the platform payload in whole seconds, and a second means the same thing on TikTok and on Instagram. It is not modelled, not rounded at source and not inferred. The judgement on this page is the label attached to each video, which decides which pile a duration lands in.
- Do these times include Instagram Reels?
- Yes. 250 of the 1,100 videos are Instagram and 850 are TikTok. The split is uneven by format: 42 of the 100 promos are Instagram, against 10 of the 100 listicles. TikTok medians run longer in 8 of the 11 formats, so the platform mix inside a format moves its median and is printed alongside it.
- Why is a listicle almost three times as long as a promo?
- We can say what the gap is, not what causes it. The medians are 82 seconds and 30. Both formats run wide: the middle 80% of listicles falls between 15 and 179 seconds, and the middle 80% of promos between 7 and 99. A listicle enumerates items and a promo asserts one thing, so the running times differ in the direction the formats imply. This draw tests nothing about why.
- What is the 'other' format?
- The residual bin in the classifier's vocabulary, not a format anyone sets out to make. Its median of 15 seconds is the shortest of the 11, and 48 of its 100 videos ran under 15 seconds. Read it as where short clips with nothing to classify end up, and never as a length worth copying.
- Can I see the running time of a competitor's videos?
- Yes. Duration sits on every organic video record alongside its platform, its publish date and its labels. The organic view filters by content type. Timing one competitor's last 20 posts is a different question from this page, which reports 100 videos per format across the whole library.