Buffer's 2.7 Million-Reel Study: 90 Seconds to 3 Minutes Wins Reach

Reels between 90 seconds and three minutes reached 4–7% more unique accounts than an account's typical reel — the strongest length signal in Buffer's analysis of 2,730,166 Instagram Reels published through its platform between January 1 and August 15, 2026.

The data science team at Buffer (researchers Steve and Iggy) started from 3,780,620 reels with usable duration data, then narrowed the set for fair comparison. The surprise: super-short clips didn't win. Reels over three minutes got the most engagement, but paid for it in distribution — 49.3% below baseline reach and 45.9% fewer views.

The reason sits in platform policy, not algorithm mystery. Instagram caps recommendation eligibility at three minutes. Upload anything longer and the app warns: "Reels over 3 minutes won't be recommended to new audiences." Followers can still see it, but it won't surface in the Reels tab, Explore, or non-followers' feeds. Reels can technically run up to 20 minutes, though some accounts still see 90-second or three-minute caps from the gradual rollout.

What does the data say by length bucket?

Buffer compared each account to itself, requiring at least three reels per bucket and two buckets overall, which keeps giant followings from skewing results. Reach figures versus each account's baseline:

• 90–179 seconds: +7.4%
• 30–59 seconds: +2.7%
• 60–89 seconds: +1.6%
• 15–29 seconds: −0.9%
• Under 15 seconds: −1.5%
• 180 seconds or longer: −49.3%

That +7.4% is an average. The median account saw closer to 4%, meaning a minority of accounts pulled the number up. Most creators landed nearer the bottom of the 4–7% range.

What does this change for practitioners?

Two practical rules fall out of the data. First, if growth is the goal, 90 seconds to three minutes is the sweet spot, and three minutes is the hard stop for recommendation eligibility. Second, if nurturing existing followers matters more, long reels earn an engagement rate 2.65 percentage points above baseline (1.1 points for the typical account) — the highest of any bucket, from roughly half the audience.

Below 90 seconds, length barely registers. Every bucket from under 15 seconds to 89 seconds landed within about three percent of baseline in both directions. The hook, topic, editing, and value matter far more than the choice between 30 and 60 seconds.

For testing, Buffer recommends trial reels (a format shown to non-followers first), running two versions with identical editing, hook, and music, then waiting 24 hours for a clear read. Trial reels still work for uploads over three minutes, per the author's own test.

Two workarounds exist for long content: split it into a series of two-minute parts, both recommendation-eligible, or post a 90-second teaser pointing to the full version on your profile or Stories.

Head of Instagram Adam Mosseri has said plainly that Instagram isn't trying to be YouTube, so the three-minute recommendation cap looks durable. Cross-posters get one convenience: the three-minute limit also works on TikTok (10 minutes in-app, 60 uploaded) and YouTube Shorts (3 minutes, after which videos publish as regular uploads).

On methodology: views and reach were log-transformed to stop single viral posts dominating averages, and engagement rate was capped at the 99th percentile, around 54.8%, to filter broken denominators as extreme as 827,200%. Buffer found a similar reach-versus-engagement trade-off in earlier carousel research, suggesting the pattern runs deeper than one format.