We split 3,100 Reels into quarters by reach, and most of the advice did not survive
We split 3,100 Reels into quarters by reach. Length, engagement rate and caption size barely separate them. Queried 28 September 2026.
What the numbers say
We took 3,100 Reels, divided each one by its own account follower count so that reach means reach rather than audience size, then split them into four equal groups. The things people are told to optimise barely move. Average duration runs 60.4, 59.9, 59.8 and 57.1 seconds from the best quarter to the weakest, a spread of three seconds. Median engagement rate runs 0.0443 down to 0.0313, so the best quarter is about a third more engaging, not a multiple. Median caption length runs 266, 220, 220 and 210 characters, meaning the Reels that travelled furthest had longer captions, not shorter. The one figure that separates them clearly runs the wrong way from the advice: comments as a share of engagement run 0.012 in the best quarter against 0.046 in the weakest, nearly four times lower.
The comment share figure is a symptom, not a lever. A Reel that travels far beyond its own audience collects a great many passive likes from strangers, which pushes the comment share down. Chasing fewer comments would be acting on the arithmetic rather than the audience. What the rest of it says is more useful: duration, engagement rate and caption length do not tell you which Reels travelled. If a checklist promises virality through any of those three, this corpus does not support it.
Sample of 3,100, measured on 2026-09-28. Scrptly production database, queried 2026-09-28. 3,100 Reels with more than 1,000 views, at least one like, and an account with a recorded follower count above 1,000, split into four equal groups by views per follower.
The measurement
All 3,100 Reels split into four equal groups by views per follower, best quarter first, on four measures people commonly try to optimise. Read the columns downwards: duration moves by three seconds across the whole range, engagement rate by about a third, caption length by 56 characters, and comment share by nearly four times, in the opposite direction to the advice.
| Band | Reels measured | Average seconds | Median engagement rate | Median comment share | Median caption characters |
|---|---|---|---|---|---|
| Best quarter by reach | 775 | 60.4 | 0.0443 | 0.012 | 266 |
| Third quarter | 775 | 59.9 | 0.0422 | 0.025 | 220 |
| Second quarter | 775 | 59.8 | 0.0369 | 0.030 | 220 |
| Weakest quarter by reach | 775 | 57.1 | 0.0313 | 0.046 | 210 |
One Reel, beat by beat
One Reel from the top of the reach distribution: @thelittlejoysmama, six seconds, 2,517,037 views, five beats. Six seconds and no spoken script, which is worth sitting with given how much virality advice concerns what to say.
| At | What happens |
|---|---|
| 0:00 to 0:01 | Enter the scene: walk into frame, arrive home, approach the pause spot |
| 0:01 to 0:03 | Take the pause: sit down, stop moving, show a tired expression |
| 0:03 to 0:04 | Hold the stillness: let the exhaustion breathe, minimal movement |
| 0:04 to 0:05 | Accept reality: stand up, shift the expression to resignation |
| 0:05 to 0:06 | Exit to duty: walk out of frame toward the next task |
What to do next
- Spend the effort on the first line. It is the only script element in this data with a large effect that survives correcting for account size.
- Check the slot before you post. Weekend mornings carried 0.87 views per follower against 0.39 for the weekday afternoon where most people post.
- Stop cutting to hit a duration. Three seconds separated the best quarter from the weakest, 60.4 against 57.1, and the weakest quarter was the shorter one.
- Treat a low comment share as a sign your Reel left its own audience, not as a problem to fix.