The right length for an Instagram Reel is the shortest duration that still delivers your payoff in full — which means it changes with the goal, not with the calendar. As of July 2026, Reels run up to three minutes, and the reflex of "keep everything under fifteen seconds" is no longer automatically correct. Instagram judges a Reel by how much of itself it gets watched, so a tightly edited 80-second Reel that holds people beats a 12-second one nobody bothers to finish.
Why "shorter is better" stopped being true
That advice is a fossil from the era when Reels were capped first at 15 seconds, then 30, then 90. When the ceiling is low, the only lever you have is making the idea fit. Now the ceiling is high enough that duration is a real choice, and the constraint has moved from the platform to your audience's patience.
What has not changed is the scoring. Reels ranking leans on watch-time signals — how long people stay, whether they re-watch, whether they send it on — rather than on raw seconds. Our breakdown of how the Instagram Reels algorithm ranks videos goes through those signals in detail, and the short version matters here: length is not a ranking factor. It is the denominator underneath the ranking factors.
That distinction is the whole game. If you are fuzzy on how watch-time and completion relate to each other, watch time and retention explained is worth ten minutes before you script anything.
Length by goal
Pick the goal before you pick the duration. Four common goals, four working bands:
| Goal | Working length | Why it works |
|---|---|---|
| Reach and discovery | Under 15s, loopable | Cheap to finish, easy to re-watch, high completion |
| Conversion or a specific ask | 20–45s | Long enough to make the case, short enough that the CTA lands |
| Nurture and authority | 60–90s | Room for setup, substance, payoff — the trust-building band |
| Deep teaching | 90s–3min | Maximum watch-time per viewer, but only if every beat earns it |
Treat these as starting defaults, not laws. They are where I begin when there is no data yet; the data replaces them within a few weeks.
Reach: under 15 seconds, built to loop
When the job is getting in front of strangers, you want the highest completion rate you can engineer, and the cheapest way to get it is to ask for very little. A clip that ends before anyone decides to swipe gets finished by most of the people it reaches. Design the last frame to flow back into the first and you collect re-watches without the viewer consciously choosing to replay.
This band suits single-idea clips with an instant visual hook, trend formats the audience already recognises, satisfying loops, and blunt opinions that bait a comment. What it cannot do is teach. You will build awareness here and very little else.
Conversion: 20 to 45 seconds
A freelance social media manager running a dental client's "new patient offer" Reel does not need three minutes. She needs to show the problem, show the result, and make one clear ask before attention decays. Front-load the value, keep the ask singular, and end while people are still watching. The classic failure is burying a call to action at second 50 of a Reel that lost half its viewers by second 20.
Nurture: 60 to 90 seconds
This is the band most small accounts underuse. Sixty to ninety seconds gives you a proper setup, a middle with real substance, and a resolution — the shape that turns a passing viewer into someone who follows. A small agency explaining "why your last three posts flopped" cannot do it in twelve seconds without sounding glib. Given eighty, it sounds like expertise.
The catch is that longer only works with retention editing. Cut the dead beats, change the visual every few seconds, and plant open loops ("the third mistake is the one everyone makes"). Our guide to making Reels people watch to the end covers the pattern interrupts and pacing that keep ninety seconds feeling like thirty. Without those moves, a 90-second Reel is just a 20-second Reel with seventy seconds of drag attached.
The three-minute ceiling: available, rarely correct
Three minutes exists for genuine depth — a full walkthrough, a multi-step tutorial, a story that needs room to land. Availability is not a recommendation. Most accounts that stretch to three minutes are padding, and padding shows up immediately as a sagging retention curve. If you cannot defend every thirty-second block out loud, cut it into two Reels instead.
Completion is scored relative to length
Here is the mechanic that makes "length by goal" work rather than "length by fashion."
Imagine two Reels from the same bakery. The first is 10 seconds of dough being folded; nearly everyone finishes it, and a chunk of viewers loop it twice. The second is 100 seconds explaining why supermarket sourdough is a lie; most people leave around the 30-second mark. The short one has the better completion rate. The long one may still have produced more total watch-time per viewer — thirty seconds of attention versus ten.
Both signals count, and they pull in different directions. That is exactly why the goal has to come first. Chasing completion rate pushes you short. Chasing watch-time pushes you long. Chasing both at once, with no stated goal, is how creators end up making 45-second videos that are too long to loop and too short to teach — the dead zone where a lot of mediocre Reels live.
The practical rule: decide which signal you are buying with this Reel, then set the length to buy it.
Let your own retention data set your default
Folklore gives you a starting number. Your analytics give you the right one. Instagram's per-Reel insights show where viewers drop, and three shapes tell you almost everything:
- A cliff in the first two seconds. The hook is failing. Length is not your problem — do not shorten the video, fix the opening frame.
- A slow bleed through the middle. The pacing is too slow for the content. Tighten the edit, or make the next one shorter.
- A flat line to the end. You had more runway than you used. Your audience wanted more. Test a longer cut.
Check this per Reel, not per month. Within three or four weeks you will know whether your particular audience rewards 12-second loops or sits happily through two minutes, and that beats any benchmark from a listicle.
A note on tooling, since people ask: SocialKit does not trim, reframe, or auto-cut your video — you upload the cut you made. What it does give you is the scheduling layer and post analytics around it, so you can queue a deliberate mix of lengths across a fortnight and compare how each performed, rather than posting on impulse and forgetting to look. The second-by-second retention curve still lives in Instagram's own insights; use both.
Length is a format decision before it is an algorithm decision
A talking-head tip is naturally short. A step-by-step tutorial needs room. A before-and-after transformation is over the moment the reveal lands. Choose the format that fits the idea, and the length usually chooses itself.
Whatever you land on, shoot to spec so the frame survives. A Reel that letterboxes or crops the on-screen text loses people in the first second, which sabotages retention before duration ever matters — our social media image and video size reference has the current dimensions. If you want the full production sequence around all this, the practical Reels workflow guide walks through hook, cuts, caption, cover, and publishing.
One cross-posting warning: the same footage often wants a different cut per platform. TikTok audiences tolerate a longer setup than Reels viewers generally do, and the goal-to-length mapping differs enough to matter — the TikTok video length guide is the sibling to this piece. Re-cut rather than re-upload where it counts.
Start here
- Write the goal at the top of the script. Reach, conversion, nurture, or teaching. One of them, not all four.
- Take the default length for that goal from the table above.
- Draft the shortest version that keeps the payoff intact, then only add time you can defend out loud.
- Cut every second that neither advances the idea nor resets attention.
- Publish and check the retention curve within 48 hours. Cliff, bleed, or flat line — each points at a different fix.
- After a month, replace the defaults with your own numbers. Your account's answer is the only one that counts.
If you want to run that test methodically, SocialKit's calendar lets you plan a spread of Reel lengths across the coming weeks alongside the rest of your posting and review how each one did — flat pricing from €29/month Solo (€17.40/month billed annually as of July 2026), with a 7-day free trial. The tooling will not tell you how long your Reels should be. Your watch-time data will, once you are posting consistently enough for the pattern to show.