Skip to content
All posts

How does the Instagram algorithm work in 2026?

By Waris Hussain
Last updated 15 November 2026 · 6 min read

In 2026, Instagram ranks posts mainly on three signals: watch time, likes and sends. It tests public posts with non-followers first, limits recommendations for accounts that mostly repost, and caps hashtags at five. To read your own performance, divide each number by reach, as in sends per reach.

Most "algorithm" advice online is recycled from 2021. Everything here comes from what Instagram has said, with dates, plus large studies that state their sample size. I update it when Instagram changes something.

Is there one Instagram algorithm?

No. Instagram runs separate ranking systems for Feed, Reels, Stories, Explore and Search, and each one predicts something slightly different. What they share is the goal: show each person the posts they're most likely to value.

For a creator, the useful split is between followers and non-followers. Your followers see you in Feed and Stories. Everyone else finds you through Reels, Explore and recommendations. Growth comes from the second group, and that group is ranked on slightly different signals, which is why a post can do well with your followers and still go nowhere.

Which signals matter most in 2026?

Instagram's head, Adam Mosseri, named three in January 2025 (Social Media Today, 22 Jan 2025): "The top three signals that matter most for ranking are watch time, likes and sends." He has kept pointing creators back to them since.

Signal What it measures Weighs slightly more with
Watch time How long people keep watching Everyone
Likes Whether people enjoyed it Followers
Sends Whether people share it in DMs Non-followers

Sends are the one most creators underrate. In a Stories Q&A on 23 October 2026, Mosseri said comments count and that Instagram focuses a lot more on finding posts you'd send to a friend (ZoomSphere, Oct 2026). I wrote a full post on that: Sends vs comments.

What changed in 2026?

Five changes matter for small creators:

Date Change What to do
18 Dec 2025 Hashtags capped at 5 per post or reel (Social Media Today) Use up to 5 specific tags, chosen for search
2 Apr 2026 Trial Reels can be scheduled (Social Media Today) Test hooks on non-followers first. See Trial Reels
30 Apr 2026 Originality rules extended from Reels to photos and carousels: accounts that mostly post others' content lose recommendation eligibility (Instagram) Make most of your last 30 days original. See originality rules
Sep 2026 Mosseri explains four ranking principles: originality, recency, helping unfamiliar creators break through, and bigger models weighing more signals (Social Media Today, 24 Sep 2026) Write every hook for a stranger
28 Oct 2026 Linking Reels into a series removed (Social Media Today, 15 Oct 2026) Build series without the "part 2" link. See series

Instagram changes features often, so check the date on each row against what you see in the app.

How does Instagram test a new post?

In his September 2026 explainer, Mosseri described showing posts from small creators who post publicly to people who don't follow them, to see how they respond (Social Media Today, 24 Sep 2026). So some of the first people to see a public reel may be strangers. If they skip, the post stalls. If they watch, like and send, it gets shown to more people.

That has two practical effects. First, your hook has to work for someone who has never heard of you (see how to write a reel hook). Second, follower count matters less than most people think. In his Colin and Samir interview, Mosseri said a brand-new account can get millions of views on its first post (Net Influencer, 28 Oct 2026).

Does Instagram rank AI content lower?

Probably, a little. In the same Colin and Samir interview, reported on 28 October 2026, Mosseri said Instagram ranks AI creators the same way as other creators, and that its systems probably don't yet rank AI content as well as other content (Net Influencer). He also said Instagram is starting to ask AI-generated characters to label themselves.

I'm building an animated coach, Cleo, so this one is personal. Our answer is to keep Cleo's posts grounded in things only we can show: teardowns, numbers from accounts we have permission to use, and our own analysis.

How should you measure what's working?

Track rates per reach. In September 2026, Mosseri told creators who want more reach to judge posts by engagement rates and give total views and likes less weight (Social Media Today). In October he named sends per reach as the number to watch for reaching non-followers, and said Instagram is bringing it into Insights (Net Influencer).

  • Sends per reach (sends ÷ accounts reached)
  • Likes per reach
  • Average watch time and skip rate

A reel with ten times the views can still have fewer sends per reach than a smaller one, which means strangers valued the smaller one more. I break down where to find each number in How to read Instagram Insights like a coach.

What do the big studies say about reach in 2026?

They say reach per post is under pressure while total views are rising. Here are the largest recent studies:

Study Sample Instagram finding
Metricool 2026 Social Media Study, Dec 2025 39.8M posts from 1.06M accounts, 2025 data Post reach down 31% and Reels reach down 35% year over year
Metricool 2026 Instagram Study, Jun 2026 24.4M posts from 375,118 accounts, Jan to Feb 2026 vs 2025 Views up 26.56%, single-image reach down 21.96%, average Reels watch time up to 8.5 seconds
Buffer State of Social Media Engagement, Mar 2026 52M+ posts across 10 platforms Median Instagram engagement rate fell from about 7.3% (2024) to 5.4% (2025)
Emplifi Q3 2026 benchmarks, Oct 2026 77,038 brand Instagram profiles On Reels, shares are 9.7% of interactions and comments 3.8%

Two cautions. These samples are mostly brands and the tools' own users, so they describe the platform as a whole and may not match your account. And when Metricool also found brand content volume up 24% in the same period, a lower average per post is partly more posts splitting the same attention.

So a drop that matches these numbers may not be about you. Compare yourself to your own past rates first. If your reach fell, the reach drop check walks through it.

What does this mean for you this week?

  1. Check sends per reach on your last 10 reels.
  2. Make sure most of your last 30 days of posts are your own.
  3. Rewrite your next hook as if nobody watching knows you.
  4. Use up to 5 hashtags, chosen for search.

This is what Cleo, Creator Kitchen's content coach, does with your account: she studies every post you've made, then shows you which of these signals is holding you back, with your numbers attached. Join the waitlist.

Sources