How Long an Instagram Reel Should Actually Be.

Instagram will let you post a Reel that runs for twenty minutes.

That does not make twenty minutes a sensible length for anything.

The number that actually governs a Reel is three. Instagram’s stated rule is that it only recommends Reels under three minutes long, which means anything past that line is shown to people who already follow you and almost nobody else. The video still publishes. It still sits on your grid. It just stops being eligible for the surfaces where strangers find you.

So the short answer to how long a Reel should be is this: under three minutes if you want reach, and somewhere around a minute if you want the length that tends to perform best.

The three minute line is the only number that decides reach

Every other spec on the Reels format is a container. Three minutes is a gate.

Under it, your Reel can be pushed into the Reels tab and into Explore, in front of people who have never heard of your brand. Over it, distribution collapses back onto your existing follower graph.

That is a different product. A Reel under three minutes is an acquisition asset. A Reel over three minutes is a retention asset for an audience you already own.

Most teams treat those two things as the same post with a different runtime, and then wonder why the long one did nothing.

What Instagram actually lets you post

The published limits are less tidy than the headline suggests, and they differ depending on how the video gets into the app.

• Minimum length: around 15 seconds.
• Filming inside the Instagram camera: up to 20 minutes.
• Uploading a finished file from your camera roll: capped lower, commonly at 15 minutes.
• Recommended to non-followers: under 3 minutes, and nothing above it.

Not every account sees the same caps at the same time, because Instagram rolls these changes out in waves. If your editor is capped at a shorter length than a colleague’s, that is the rollout, not a bug in your file.

None of this changes the planning number. Whatever your app allows, three minutes is the line that decides whether the work can find anyone new.

The length brackets, and what each one is for

Think of Reels in four bands, each with a different job:

• Under 30 seconds: hooks, single ideas, reactive posts, anything that lives or dies on the first two seconds.
• 30 to 90 seconds: the workhorse band. Enough room for a setup and a payoff, still fully eligible for recommendation.
• 90 seconds to 3 minutes: deeper stories and demonstrations. Slightly lower reach in practice, stronger watch time from people who care.
• Over 3 minutes: tutorials, interviews and evergreen reference material you will drive traffic to yourself, because Instagram will not do it for you.

The fourth band is legitimate. It is just not a growth band, and it should never be the default.

Where the numbers cluster

Every large study of Reels performance lands in roughly the same place. Engagement peaks somewhere around the 45 to 90 second range, and the short end does respectably.

The interesting part is the bottom of the table. Reels over three minutes come in worst on both engagement and median views, and by a wide margin rather than a nose.

That is not an audience taste finding. It is the recommendation gate showing up in the numbers. Those videos were not judged and rejected, they were never shown to anyone who could have judged them.

Which is why a long Reel with a good watch-through rate can still look like a failure. The rate is fine. The denominator never grew.

Adam Mosseri’s own framing is that there is no perfect length and the goal is to keep people watching. That is true inside the gate. It is not a licence to cross it.

Length is a distribution decision before it is a creative one

This is the part most content teams have backwards. They cut the video to whatever the story needs, then check whether it fits.

Length is not the output of an edit. It is an input to the brief, in the same way that aspect ratio and platform are inputs. Deciding to make a four minute Reel is deciding, in advance, that this piece will not reach anyone new.

Sometimes that is the right call. It just needs to be a call, made by someone who knows what it costs, and not something discovered in the edit at four in the afternoon.

That is the same discipline behind any working short form video strategy: the format constraints are agreed before the camera comes out, not negotiated afterwards.

What most teams get wrong about the twenty minute ceiling

A raised ceiling reads like permission. It is not. It is storage.

Three failures follow from reading it as permission:

1. Long-form content gets dumped onto Reels unedited, because it technically fits, and then reaches only the people who would have seen it anyway.
2. The team starts judging performance on the wrong baseline, comparing a four minute Reel’s views against a 50 second one and concluding the topic failed.
3. Editing standards slip, because a hard cap was doing the work that editorial judgement should have been doing.

The last one is the expensive failure. A tight limit forces choices. Remove it and the average edit gets looser, and looser edits lose people early regardless of how long the video runs.

If you have long assets worth using, the answer is to cut them properly rather than to post them whole. That is why NBK treats video production and editing as a system built around a platform’s constraints rather than one that discovers them late.

When a Reel over three minutes is the right call

There is a narrow set of cases where crossing the line makes sense:

• A tutorial or walkthrough your existing customers will actively search for and return to.
• A founder interview or long explainer you plan to distribute yourself, through email, a newsletter, a website page or a pinned post.
• Content whose value is completeness, where cutting it into parts would destroy the point.

In all three cases you are using Instagram as a host, not as a distribution channel. That is a fair use of the platform, provided you have another way of getting people to the video.

If you cannot name that other way, the Reel needs to come in under three minutes.

How to decide length before you shoot

Ask three questions at the brief stage, in this order.

First, who is this for: people who already follow us, or people who do not? That single answer sets your ceiling.

Second, what is the one thing this video has to land? If you cannot state it in a sentence, extra minutes will not rescue it.

Third, what is the shortest runtime that can carry that one thing honestly? Start from the answer and add only what earns its place.

A brief that arrives with a target length and a reason for it produces a very different edit from one that arrives with a rough idea and a twenty minute allowance.

Checks before you publish anything over three minutes

If a Reel is going out long, run it past these first:

• Is there a genuine reason this cannot be cut to under three minutes, beyond effort?
• Do we have a distribution plan that does not depend on Instagram recommending it?
• Have we accepted, in advance, that its view count will look poor next to our short Reels?
• Is there a shorter cut of the same material going out as well, to do the acquisition work?
• Will the person reading the monthly report understand why this one looks different?

The last check matters more than it sounds. Unexplained outliers in a report get treated as failures, and good long-form work gets killed by a chart nobody contextualised.

What the rollout still does not tell you

Instagram has not published a full breakdown of how length maps to distribution below the three minute line, and anyone claiming a precise curve is guessing.

What is documented is the gate itself. What is measured, by third parties on large samples, is that the middle of the range performs best in aggregate.

Aggregate is not your account. Your own retention curves are the better evidence, and they are the only evidence that reflects your audience, your subject matter and your edit style.

Treat the three minute rule as a hard constraint and everything below it as a hypothesis to test.

What good looks like

A team with this solved has a default length written down, usually somewhere between 30 and 90 seconds, and it appears on the brief before anyone books a shoot.

Anything over three minutes needs a named reason and a named distribution route. It is a decision with an owner, not an accident of the edit.

Long assets are planned as sources, not as posts. One shoot produces a short cut for reach and, where it is justified, a longer cut for the people already paying attention.

And performance gets read by cohort. Short Reels are judged on new-audience reach, long Reels on completion and saves among followers, so the two never get compared on a single number.

How NBK thinks about video length

Length is an operations question wearing a creative costume. The failure is almost never that someone chose the wrong runtime, it is that nobody chose one at all.

NBK’s team has shipped video at publisher scale, tens of thousands of posts across platforms where the format rules change constantly. What holds across all of it is that the constraint belongs in the brief, agreed by the person accountable for reach, before production spends money.

That is the same principle behind cadence, approvals and reporting. Decisions made early and written down are cheap. The same decisions made in the edit, or not made at all, are what turn a busy content team into an ineffective one.

If your Reels are inconsistent in length with no rule behind it, the fix is a standard, not a better editor.

Next step

Pull your last thirty Reels, group them by runtime, and compare reach from non-followers across the groups. If everything over three minutes sits flat, you have found a distribution ceiling rather than a content problem, and you can start reading skip rate and share rate on the Reels that actually got a fair shot.

If your social output feels busy but not effective, start with an NBK audit.