Instagram does not have one algorithm — it runs a separate ranking system for Feed, Reels, Stories, Explore, and Search, each optimizing for a different behavior. Three signals cut across all of them (watch time, likes per reach, and sends per reach), the DM send is the loudest of the three, and in 2026 a new layer sits on top of everything: an originality standard that demotes reposted and aggregated content and rewards net-new, first-party work. This guide maps how each surface ranks, what "connected vs. unconnected reach" means for your strategy, and the concrete moves that earn distribution now.
The first strategic mistake is talking about the Instagram algorithm as one thing. Instagram has been explicit for years, through Mosseri and its own documentation, that it runs a different ranking system for each surface of the app: one for Feed, one for Reels, one for Stories, one for Explore, and one for Search. Each was built to predict a different action, so each weighs signals differently. A Reel engineered to travel through recommendations can barely appear in your own followers' Feed; a carousel that your audience saves and revisits may never surface on Explore. Treating the app as a single ranking box leads you to optimize one number and wonder why another collapses.
So the real 2026 strategy is not "beat the algorithm." It is to decide, per post, which surface you are trying to win, and then feed that surface the behavior its ranking system rewards. This guide maps those systems. It is the mechanics companion to two others already on the site: the [Instagram marketing strategy](/guides/instagram-marketing-strategy) guide covers the format-and-cadence playbook, and [Instagram trends 2026](/guides/instagram-trends-2026) covers what shifted this year. Here the focus is narrower and deeper — how ranking itself works, and the moves that follow from understanding it.
Underneath the per-surface differences, Mosseri has consistently named the same three signals as the ones that matter most across the app: watch time, likes per reach, and sends per reach. "Per reach" is the crucial phrase — Instagram cares about the rate at which the people who saw your post engaged, not the raw count, which is what lets a small account outrank a large one. A post shown to 1,000 people that earns 200 saves is a stronger signal than one shown to 100,000 that earns 500. Optimizing for reach-normalized engagement is why chasing followers for their own sake is a weak strategy: the algorithm grades the quality of the response, not the size of the audience.
Watch time is the workhorse signal for video. Instagram tracks total seconds watched, the percentage of a Reel completed, and whether people rewatch — and the first few seconds are where distribution is won or lost, because a fast drop-off tells the system the content is not worth pushing. This is why the hook matters more than the payoff for reach: a Reel that holds viewers to the end gets shown to more non-followers than a longer one that loses half its audience early. Likes per reach is the lightest of the three, a baseline positive vote. The one that has quietly become the center of Instagram strategy is the send.
Mosseri has called sends per reach one of the biggest signals Instagram uses in ranking, and it is widely reported as carrying several times the weight of a like for Reels distribution. The reason is mechanical, not sentimental. When someone taps the paper-plane icon and forwards your Reel into a friend's DMs, they hand Instagram a free, high-confidence recommendation to a specific person who does not follow you — which is exactly the audience expansion the recommendation engine is built to manufacture. A like is a private signal that lives on your post; a send is distribution that leaves it. In 2026, "would someone send this to a friend?" is a more useful question than "would someone like this?" — and it changes what you make. A stat surprising enough to forward to a colleague, a tip specific enough to send to a teammate, a joke one friend has to see: that is send-optimized content, and it out-travels content built only to be admired.
Saves work on the same logic as a value signal, which is why the carousel — the format people save and return to — carries so much weight on the engagement surfaces even though Reels win raw reach. If you take one thing from the signal layer, take this: design for the actions that move your content to new people (sends) or mark it as worth keeping (saves), and treat likes as a byproduct rather than a target.
Instagram's own framing splits your distribution into two pools, and understanding the split is what turns the signal list into a strategy. Connected reach is the people who already follow you, seeing your posts and Stories because of your existing relationship. Unconnected reach is everyone else — non-followers who encounter you through recommendations in Reels, Explore, and the suggested posts now seeded into Feed. These two pools are ranked by different systems and respond to different inputs.
Connected reach is governed largely by relationship history: how often a given follower has interacted with you, DMed you, watched your Stories, tapped your profile. It is relatively stable and hard to spike, because it is a function of accumulated behavior between two accounts. Unconnected reach is governed by content-quality signals — watch time, sends, saves — evaluated fresh on each post with little regard for who made it. That is why a single Reel from a small account can reach millions while the next one barely moves: unconnected reach is decided post-by-post on merit, not on your follower count. The strategic consequence is direct. Growth comes almost entirely from unconnected reach, so the content you want to grow with must be engineered for recommendation signals. Retention and relationship come from connected reach, which you protect with consistency and genuine interaction, not with hooks. Trying to make every post do both jobs is the most common way strategies stall.
With the signals and the two reach pools in hand, here is how each individual system ranks — and the move that follows from each.
The Feed system predicts a small set of actions: how likely you are to spend a few seconds on a post, comment, like, save, and tap the profile. It weighs your interaction history with the poster heavily, so Feed is where connected reach lives — your existing relationships decide most of what surfaces. Increasingly, though, Feed also injects recommended posts from accounts you do not follow, ranked on content signals. The move: for your followers, post things that earn a comment or a save (they deepen the relationship history that keeps you in their Feed); for the recommended slots, the same unconnected-reach rules as Reels apply.
Reels is the primary unconnected-reach engine and the most content-merit-driven surface. It ranks on watch time, sends, likes, and saves, largely independent of who you are, then shows promising Reels to progressively larger pools of non-followers. Sends are the heaviest signal here specifically because Reels is where Instagram is trying to expand your audience. The move: front-load the hook, keep it tight enough to hold completion, and build for the forward. This is the surface where a single post can change your account's trajectory, so it is where experimentation pays.
Stories is pure connected reach — it is shown to followers, ranked by how likely you are to tap in, reply via DM, or move on. It rewards frequency of viewing and the strength of the offline-style relationship. It will not grow you, and that is fine: its job is retention and intimacy. The move: use Stories to maintain relationship history with existing followers (which feeds your Feed ranking with them), not to chase reach it structurally cannot deliver.
Explore is a recommendation grid built almost entirely from accounts a user does not follow, ranked on post popularity — likes, comments, shares, saves, and how fast they accumulate — plus the user's own Explore behavior. It is unconnected reach for image and carousel content the way Reels is for video. The move: make grid-stopping, save-worthy posts, because velocity of engagement in the first window is what pushes a post from a few Explore impressions to many.
Search is the surface most creators underuse. Instagram became a search engine, and its Search ranking runs on keywords — the words in your captions, your name field, and your bio — far more than on hashtags. Following a hashtag was removed in December 2024, and tags no longer drive discovery on their own. The move: treat your profile and captions as SEO real estate. Put the terms your audience actually types into search where the system can read them, and stop relying on a wall of 30 tags that no longer does anything.
The signal systems above describe how Instagram ranked content going into 2026. The defining change this year is a layer that sits on top of all of them and asks a different question — not "is this good?" but "is this yours?" It arrived in two steps. On December 31, 2025, Mosseri published a year-end memo (a 20-slide carousel) laying out Instagram's priorities, and the headline was authenticity: as AI makes polished content trivial to fake at scale, Instagram would prioritize raw, real, human content. The line that traveled was the reframing of the bar itself — from "can you create" to "can you make something that only you could create."
The memo was direction; the enforcement came on April 30, 2026, when Instagram announced that accounts primarily reposting other people's work would no longer be eligible for recommendations across the app, including Feed and the central Discover feed. Crucially, the policy extended a rule that had mostly applied to Reels out to photos and carousels too — so aggregator accounts across every format lost their access to unconnected reach. Followers who deliberately follow an aggregator still see its posts, but the recommendation faucet — the growth engine — gets shut off. Instagram defined original content as work a creator wholly made or that reflects their unique perspective, or third-party content that has been substantially edited (their example: a meme that transforms someone else's image by adding genuine humor, commentary, or a fresh take). What does not qualify is telling: low-effort edits like added watermarks or speed changes, and screenshots of others' posts even when the original creator is credited.
For strategy, the originality reset means the source of your content is now itself a ranking input. Two accounts can post the identical clip and get opposite distribution — the one that made it reaches non-followers, the one that reposted it does not. This is the single biggest reason a repost-and-recycle approach to Instagram is a dead end in 2026, and it rewards, structurally, anyone producing net-new first-party content. It also raises the production bar in the same breath as raising the authenticity bar, which is the tension the rest of your strategy has to resolve.
Pulling the mechanics into moves: First, pick a surface per post and build for its system — a Reel engineered for sends is a different object than a carousel engineered for saves, and pretending one post serves both is why reach flatlines. Second, treat unconnected reach as your growth budget and connected reach as your retention budget; spend Reels and Explore effort on the former, Stories and relationship effort on the latter. Third, optimize for the send above the like — the forward is the highest-leverage action in the system, so make content specific and useful enough that one viewer has to pass it to one other person. Fourth, win the first three seconds of every video, because watch-time drop-off there caps everything downstream. Fifth, move discovery effort from hashtags to search keywords in captions, name, and bio.
Sixth, and now foundational: clear the originality bar. Post content you actually made, in a register that reads as human — the selfie-cam talking head, the original audio, the specific point of view — because both the classifier and the audience are discounting recycled, watermarked, and generically-polished material. The 2026 algorithm is not asking you to game a signal; it is asking you to produce a real, format-rich, first-party stream at a cadence. That is a coherent strategy and an honest one. It is also a serious production problem, which is where most accounts get stuck — the plan is right and the throughput is the wall.
Every conclusion above converges on one operational demand: produce net-new, first-party content, sized differently for each surface, at a cadence, without it sliding into generic AI polish. That is a volume-and-originality problem at the same time, and it is precisely the gap a generation-and-publishing engine is built to close. The distinction that matters most here is that Kompozy generates original content rather than reposting it — which puts it on the right side of the 2026 originality reset by construction. A repurpose-only tool that recycles one clip into re-uploads is exactly what the April 2026 policy demotes; an engine that creates net-new video, images, carousels, and text produces the first-party work the classifier is designed to reward.
The surface map is where the breadth pays off, because different ranking systems want different objects. For the unconnected-reach engines, Persona Shorts generates authentic talking-head Reels in the human, selfie-cam register that now out-distributes studio gloss, and Clipped Shorts cuts your real long-form footage into vertical Reels so most of the frame is genuine video the originality signal reads as first-party. For the save-and-Explore surfaces, Carousel Posts and Persona Photos and Quote Graphics produce the grid-stopping, save-worthy formats — rendered brand-exact through HyperFrames so the slides carry your real palette and type. Text Posts feed the conversational surface and give captions the search keywords that Search ranking now runs on. Producing across these from one brief is also the structural defense against a feed of near-identical posts, which the recommendation systems discount.
The consistency the algorithm rewards is enforced by the engine, not left to willpower: the Persona Brief governs voice and banned words across every caption and slide so the cadence does not flatten into the default LLM register audiences scroll past, and Gemini face-lock keeps the persona's face identical across every video and image so the brand stays recognizable — which is what both viewers and the ranking systems reward. Then Kompozy schedules and publishes the batch to Instagram natively, alongside the eight other platforms where the same content earns extra reach, behind a per-post review pipeline so a human still gates what ships. Be clear on the boundary, because honesty converts: Kompozy does not manufacture watch time, send the DMs, or fake the engagement that drives distribution — those come from the content being genuinely good and genuinely yours. What it removes is the production ceiling that stops most creators from feeding every surface, in original first-party content, at the volume 2026 ranking now requires.
Instagram in 2026 rewards a specific and narrow band of behavior, and the strategy falls out of the mechanics once you stop treating the app as one algorithm. It is several ranking systems, one per surface, so you win by matching content to the surface you are targeting. Three signals cut across all of them — watch time, likes per reach, and sends per reach — with the send the loudest because it buys unconnected reach. Growth lives in unconnected reach and is decided post-by-post on merit; retention lives in connected reach and is built on relationship. And sitting over all of it is the originality reset: reposted and aggregated content lost its access to recommendations, and net-new, human, first-party work is what the system now promotes. The moves are clear; the constraint is producing enough original, format-right content to run them. Solving that production constraint is what frees you to play the strategy at all.
There is no single algorithm. Instagram runs a distinct ranking system for each surface — Feed, Reels, Stories, Explore, and Search — and each optimizes for a different action, so content that wins on one surface can barely show up on another. Across all of them, Instagram head Adam Mosseri has named the three signals that matter most: watch time, likes per reach, and sends per reach (DM shares). Sends are the heaviest, because forwarding a post to a friend hands Instagram a high-confidence recommendation to someone who does not follow you yet. On top of the ranking systems, a 2026 originality standard demotes reposted and aggregated content and favors net-new, first-party work.
The DM send. Mosseri has repeatedly called sends per reach one of the biggest signals Instagram uses, and it is widely reported as carrying several times the weight of a like for Reels distribution. The logic is that a like is a private nod, while a send pushes your content to a non-follower — exactly the audience expansion the recommendation system exists to produce. Watch time and likes per reach also matter, but the strategic priority is making content people forward, not just tap a heart on.
Connected reach is the people who already follow you seeing your content in their Feed and Stories; unconnected reach is non-followers seeing it through recommendations — Reels, Explore, and the Feed suggestions. They are ranked by different systems. Connected reach is driven by your relationship history with each follower, so it is relatively stable. Unconnected reach is driven by content-quality signals like watch time and sends, which is why a single Reel can explode to millions or barely move. Growth comes from unconnected reach; retention comes from connected reach, and you optimize for each differently.
The biggest change is the originality reset. In a December 31, 2025 year-end memo, Mosseri said Instagram would prioritize raw, real human content over AI-generated and recycled material, framing the bar as shifting from "can you create" to "can you make something that only you could create." Then on April 30, 2026, Instagram announced that accounts primarily reposting others' work would no longer be eligible for recommendations across Feed and the central Discover feed — extending a policy that previously applied mainly to Reels to photos and carousels too. Original means wholly made or substantially edited; watermarks, speed changes, and screenshots with a credit do not qualify.
Barely. You can no longer follow a hashtag (that was removed in December 2024), and hashtags no longer drive follows or carry discovery on their own. Instagram became a search surface, so keyword-rich captions and a searchable name and bio do far more for discoverability than stacking tags. Use a small handful of genuinely relevant hashtags at most, and put the real effort into the words your audience actually types into search.
Instagram in 2026 is not one algorithm but several — a separate ranking system for Feed, Reels, Stories, Explore, and Search, each optimizing a different behavior. Three signals cut across all of them: watch time, likes per reach, and sends per reach, with the DM send the heaviest because it pushes content to non-followers. Distribution splits into connected reach (followers, driven by relationship history) and unconnected reach (recommendations, driven by content quality). On top of it all sits a 2026 originality standard: a December 2025 Mosseri memo and an April 30, 2026 policy demote reposted and aggregated content and reward net-new, first-party work.
Get started → · ← All guides · Compare Kompozy vs other tools