What is the Instagram algorithm?
The Instagram algorithm is a set of machine learning models that decide what each user sees in their Feed, Explore, Reels and Stories. There isn't a single algorithm — there are several, each tuned to a different surface and ranking signals. The shared goal is to keep users on Instagram longer by predicting which content each individual is most likely to engage with. Adam Mosseri, Head of Instagram, publishes annual transparency updates explaining what matters most. The 2025 update confirmed three priorities for creators: sends per reach, watch time and follows from non-followers — signals that show whether a post truly resonates beyond an existing audience.
How the Feed algorithm ranks posts
For each post a user could see, Instagram computes a "value score" using hundreds of signals, then ranks all candidates by score. Key signals include:
- Relationship: Does the user DM, comment on or visit this creator's profile regularly?
- Interest: Has the user engaged with similar topics, audio or aesthetics?
- Information about the post: Age (how recent), location, format, number of existing interactions
- Information about the creator: Account size, posting cadence, originality (Instagram penalizes reposted/aggregator content since 2024)
- Likely actions: How probable that the user will like, comment, share, save, dwell, watch to completion?
The five key actions Instagram tries to predict are: Time spent, likes, comments, saves and shares (sends). Shares and saves have grown in weight; passive likes have declined.
The Reels algorithm
Reels has a separate ranking model focused on discovery. The dominant signals:
- Watch time and replay rate — by far the biggest factor
- Sends per reach — shares to DMs are now the strongest engagement signal
- Likes, comments, saves
- Profile visits and follows from the reel
- Original audio (since the 2024 push against repurposed TikTok watermarks)
The Explore algorithm
Explore shows content from accounts the user does not yet follow. It heavily relies on collaborative filtering: "users similar to you engaged with this, so you probably will too." Creators get into Explore by producing content that drives saves and shares from a tight existing community.
What the algorithm does NOT do
- It does not shadowban accounts for using normal hashtags. (Some words and tags are restricted, but normal use is fine.)
- It does not favor business accounts over creator accounts.
- It does not bury posts that link out — but it does deprioritize them slightly relative to native content.
How to work with the algorithm in 2026
- Hook the first 2 seconds: Strong opening visuals; cliffhanger captions
- Maximize watch time: Use captions, multiple cuts, B-roll
- Design for shares: Ask "would someone DM this to a friend?" — relatable humor, useful tips, surprising stats
- Post consistently: 3–5 reels per week is the current sweet spot
- Engage early: Reply to comments in the first 60 minutes — Instagram weights early engagement heavily
- Original content wins: Repurposed TikToks with watermarks are downranked
- Use trending audio for reels: Picks up free distribution
Common myths
- "Posting at 9am is optimal": True only for some accounts. The Instagram algorithm shows posts to fans whenever they're active. Test your own audience.
- "You need 10k followers to get reach": False. Small accounts with high engagement-per-follower routinely outperform large ones on Explore.
- "Hashtags are dead": Partially true. Instagram now recommends 3–5 relevant tags instead of the historical 30.
Measuring algorithm performance
Track these signals in Insights:
- Reach from non-followers: The clearest sign the algorithm is amplifying your content
- Save rate: Saves divided by reach
- Send rate: Shares to DMs divided by reach
- Profile visits and follows from a single post: Indicates discovery is working
Stories algorithm
The Stories algorithm is the simplest. It ranks based primarily on:
- Direct interaction history: Whose stories you've reacted to or DM'd recently
- Recency: Newer stories rank higher
- Tap-through rate: How quickly you tap through that creator's stories
Practical implication: stories that ask for replies, polls or DMs train the algorithm to keep showing your stories to that user. Quiet stories progressively disappear from a user's tray.
The 2024–2026 algorithm shifts
Notable recent changes:
- Reduced reliance on hashtags: Down from 30 to a recommended 3–5
- Originality boost: Native content now significantly outranks aggregator reposts
- Smaller creator advantage: Instagram explicitly stated it shows users more content from smaller, unfollowed creators
- Topic-based distribution: Posts increasingly distributed to non-followers interested in the topic
- Carousel re-engagement bonus: Carousels swiped a second time count as additional engagement
publy.ch helps creators consistently hit these signals by generating hook-driven, share-worthy visuals that match what the Instagram algorithm currently rewards — without trial and error.