Between June 18 and July 23, 2026 I published 15 posts on my own Instagram account, then set every number from Insights beside Metricool's 2026 study of 24,364,803 posts across 375,118 accounts. My average Reel watch time came back at 8.71 seconds. Metricool's average across all 24.4 million posts was 8.5 seconds.

One account, 15 posts, five weeks, landing within a fifth of a second of a benchmark built from twenty-four million. I cannot take credit for that. It means the number is structural: people watch a Reel for about eight and a half seconds, then they leave. If you are a small business owner, a founder, or a nonprofit or healthcare marketer running your own social, that is the asset you are working with.

Watch: what my own Instagram audit found

The short answer

Instagram reach in 2026 is driven by three signals Instagram head Adam Mosseri named publicly: watch time, likes per reach, and sends (DM shares). Hashtags are capped at five per post and, in Mosseri's own words, "do not improve your reach." Reels are the discovery engine; carousels are the conversion engine.

My Own 15 Posts, Before Any Industry Data

The format gap was not close. My Reels averaged 182 reach; my static posts and carousels averaged 58.8. That is 3.1x, with no overlap at all: my lowest-reaching Reel pulled 113 and my highest-reaching static pulled 67. Every Reel beat every static. On views the split was 257.9 against 129.8.

8.71s my average Reel watch time, against Metricool's 8.5s
3.1x my Reels out-reached my static posts, zero overlap
0.07% my save rate: 2 saves across 2,843 views
7 LinkedIn link clicks in a year, from 18,326 impressions

The part I did not enjoy: across 2,843 total views I earned 2 saves, a save rate of 0.07%. Socialinsider's 2026 benchmarks, drawn from 35 million posts across 447,613 pages, put the median at 1 save per post for accounts my size. I was below median on the metric I had been telling myself was fine.

"My account matched a 24-million-post benchmark to within a fifth of a second. The benchmark is not a target. It is the ceiling everybody is already hitting."

What LinkedIn Told Me About Links

My LinkedIn audit produced one number worth more than the rest. Over the trailing 365 days my account had 2,149 followers and 18,326 impressions, up 55% year over year, with 799 engagements (4.36% of impressions) and 88 comments. It also had 7 link clicks all year, a rate of 0.038%.

Seven. That is the honest cost of dropping a link in the post body and hoping. If a link is your conversion path, the post has to earn the exit.

What Instagram Has Actually Said, On the Record

Most of what circulates about the algorithm is inference. A useful amount of it is not. Adam Mosseri named the three signals that matter most as watch time, likes per reach, and sends per reach, telling Social Media Today in January 2025 that "likes are slightly more important for connected content, and sends are slightly more important for unconnected content." Comments were conspicuously absent from that list.

On watch time, Mosseri told Social Media Today in February 2025: "We don't want to penalize longer videos, which is why we look at not only the percentage of a video that was watched, but also the number of seconds." Absolute seconds bank credit even when nobody finishes, which is why my 8.71 second average is something I can work with rather than a failure.

The Numbers Everyone Repeats That Are Not Real

Pulling sources for this, I tried to trace the stats in nearly every 2026 Instagram listicle. Several lead nowhere.

Do not repeat these
  • "55% of Reels views come from non-followers." Attributed vaguely to Sprout Social benchmarks, with no traceable study, sample, or date. It is not on sproutsocial.com.
  • "Reels reach 2.1x more non-followers than any other format." Attributed to Buffer, but Buffer only says Reels are "more likely to be shown to more non-followers," with no percentage.
  • Reels completion benchmarks such as "65% is good for under-15s." Sourced only to content farms crediting Sprout Social; the numbers are not on Sprout's site.
  • "A save is worth 8 to 12x a like," "aim for a 1.5% save rate," "2% in the first hour triggers Explore." No methodology, no sample, no author.
  • "Sends carry 3 to 5x more weight than likes." Appears only in secondary blogs, never in a Mosseri quote.
Use these instead
  • There is no credible published percentage for how much Reels reach comes from non-followers. Saying so beats inventing one.
  • Buffer's 4 million post analysis: Reels get 1.36x more reach than carousels and 2.25x more than single photos.
  • Socialinsider measured skip rate at 60.5% to 65.5% across all account sizes, so roughly six in ten viewers skip.
  • Socialinsider measured real save rates at 0.05% for carousels, 0.04% for Reels, 0.02% for images, one to two orders of magnitude below the popular targets.
  • Mosseri's own wording: sends are "slightly more important" for unconnected content. Slightly.

Why No Two Engagement Rate Benchmarks Agree

If you have ever compared two Instagram benchmark reports and concluded one of them was wrong, they were probably both right. They use different denominators, and most articles reprint them in one table without saying so.

SourceInstagram engagement rateDenominator
Buffer, State of Social Media Engagement 2026 (52M+ posts)about 5.46%percentage of reach
Socialinsider 2026 benchmarks (35M posts)0.45%percentage of followers
Rival IQ0.30%percentage of followers

Both are legitimate. Reach-based rates say how a post did with the people who saw it; follower-based rates say how much of your audience you activated. Pick one, label it, and only compare against benchmarks using the same denominator. Buffer's figure is down from about 7.3%, a 26% decline, and Socialinsider's is down about 24% year over year, so the direction is the same either way.

Reels Acquire, Carousels Convert

Buffer's analysis of more than 4 million posts found Reels get 1.36x more reach than carousels and 2.25x more reach than single-photo posts. Median engagement as a percentage of reach flips the order entirely: carousels 6.90%, single images 4.44%, Reels 3.31%. Buffer measured carousels driving 109% more engagement per person reached than Reels, and Metricool found carousels earn 9x more saves than single images.

That is one takeaway, not two: use Reels to be found and carousels to be remembered. Metricool also measured the cost of leaning on single images, comparing January and February 2025 with the same months in 2026: reach down 21.96%, interactions down 25.41%, engagement down 45.98%.

⚠
The hashtag caveat that matters

Metricool found posts using hashtags saw 31.70% fewer views and 33.89% fewer interactions than the platform average, and Metricool explicitly labels this correlation, not causation. Hashtags are not draining your reach. Quote it with that caveat attached or you are misquoting the study.

The Playbook I Am Running Now

One free lever: Metricool found posts asking a question earn 36.70% more comments, and a comment-focused call to action earns 202.78% more. Comments were not on Mosseri's top three, so treat it as a conversation lever, not a reach lever.

What I Still Do Not Know

Three gaps I am not going to paper over. First, no credible study establishes that saves cause reach; saves are a real Explore signal, but Mosseri's named top three were watch time, likes per reach, and sends per reach. Second, Instagram's official ranking-signals page is still dated May 31, 2023 and still lists follower count, contradicting the 2025 and 2026 Mosseri statements; where they conflict, the newer statements win. Third, Instagram has never published an optimal Reel length, only third-party data points to 30 to 60 seconds.

One encouraging finding to close on: Socialinsider's 2026 Reels analysis found accounts with 1,000 to 5,000 followers reach 9.78% of their audience, against 5.00% for accounts with 100,000 to 1 million. Smaller accounts out-reach larger ones on a percentage basis.

Prefer the business and agency angle? I wrote the version for teams and client work over at AI Powered Dahlia.

Frequently Asked Questions

No. Instagram capped hashtags at five per post in December 2025, and Adam Mosseri has said publicly that "hashtags do not improve your reach." Use five relevant hashtags in the caption for topical clarity, not for distribution.

Adam Mosseri named three: watch time, likes per reach, and sends per reach. Sends matter most for reaching people who do not follow you.

In the caption. Instagram processes the caption, alt text, and hashtags together, so first-comment placement now delays categorization without any reach benefit.

Socialinsider's 2026 analysis found 30 to 60 seconds produced the highest reach rate at 5.60%. Reels longer than 3 minutes are not recommended to non-followers at all.

Because they use different denominators. Buffer reports engagement as a percentage of reach (about 5.46%), while Socialinsider and Rival IQ report it as a percentage of followers (0.45% and 0.30%).

Sources

First-party figures come from my own Instagram Insights for 15 posts published June 18 to July 23, 2026, and my LinkedIn analytics for the trailing 365 days as of July 2026. Sample size is small by design; it is one account, reported honestly, including the parts that were below benchmark.

Dahlia Imanbay

Dahlia Imanbay

AI Strategist and Fractional CMO. I build AI systems for small businesses and mission-driven organizations, and I run social for several of them. I write honestly about what works, what does not, and what the research actually shows, including when my own numbers are unflattering. Based in the US, working globally.

Want your social to run on numbers, not guesses?

I help small businesses and mission-driven teams build content systems that are measured against real benchmarks instead of recycled statistics. If you are posting consistently and still cannot tell what is working, let's look at your actual numbers together.

Book a Conversation