How the Instagram Algorithm Works in 2026

The Instagram algorithm is not one algorithm. Instagram ranks Feed, Reels, Stories and Explore with separate systems that use different signals in a different order — and across the recommendation surfaces, Head of Instagram Adam Mosseri said on 22 January 2025 that the three signals that matter most are watch time, likes and sends.

That is the honest headline, and almost every guide on this topic flattens it into one list of ranking factors, which Instagram’s own documentation contradicts.

This page is built on a rule: every ranking claim below carries a source, a date, and a confidence label. Where Instagram has published something, you get the quote and the date it was said. Where only outside measurement exists, it is labelled as measurement. Where nothing supports a claim that is repeated everywhere, it is in the myth table with the reason it fell apart. Everything here was verified against primary sources on 24 September 2026.

The Short Version

Instagram’s only published per-surface signal ordering is from 31 May 2023 and still stands. Mosseri’s January 2025 statement names the three metrics inside those categories that matter most — watch time, likes, sends — and points you at the rate versions in Insights: average watch time, likes per reach, sends per reach. Instagram has never published a weight for any signal, so every multiplier you have read is fabricated.

The one ranking-relevant change Instagram shipped in 2026 was on originality: from 30 April 2026, accounts primarily posting content they did not create lost recommendation eligibility for photos and carousels as well as reels.

Table of Contents

• The Short Version
• What Is the Instagram Algorithm?
• The Three Signals Instagram Named — and the Words That Get Dropped
• Signal, Source, Date, Confidence
• How the Feed Algorithm Ranks
• How the Reels Algorithm Ranks
• How the Stories Algorithm Ranks
• How the Explore Algorithm Ranks
• What Changed in 2026 — and What Did Not
• Ten Instagram Algorithm Claims That Do Not Survive Checking
• What Nobody Can Tell You
• How to Measure Yourself Against These Signals
• What This Means for How You Post
• Frequently Asked Questions
• Sources and Check Dates

What Is the Instagram Algorithm?

The Instagram algorithm is a set of separate ranking systems — one per surface — that predict how likely you are to respond to a given post and order what you see accordingly. Instagram’s own framing is that there is no single algorithm but many algorithms and ranking processes, each tuned to how people use that part of the app.

There is a second split that matters more than most guides acknowledge. Mosseri describes two types of ranking: connected reach, meaning distribution to people who already follow you, and unconnected reach, meaning recommendation to people who do not. They are different problems with different signals, and a post can do well at one and badly at the other. When your reach drops, the first useful question is which of the two moved.

Everything below is organised around that structure, because acting on advice that does not name a surface is guesswork.

The Three Signals Instagram Named — and the Words That Get Dropped

On 22 January 2025, Mosseri stated that the top three signals across both types of ranking are watch time, likes and sends, and that likes matter slightly more for connected content while sends matter slightly more for unconnected content. You can read the statement on Mosseri’s own post about Instagram ranking.

Two details get dropped in almost every write-up, and both change what you would do about it.

The metrics are rates, not counts. Mosseri directed creators to the Insights versions: average watch time, likes per reach, sends per reach — each normalised by how many people saw the post. That distinction is the actionable part. A post with 400 likes from 40,000 people performed worse on the named signal than a post with 60 likes from 1,000 people, and a raw like count cannot tell you which you have.

“Slightly” is his word, and it is doing real work. He gave an ordering and one comparative qualifier, and no magnitudes. Instagram has never published a weight, a percentage or a multiplier for any ranking signal, on any surface, ever. Every figure of the form sends are worth three to five likes is an invention laid on top of a statement that says the opposite.

Signal, Source, Date, Confidence

This is the table the rest of the internet does not publish. Each row is a claim, where it came from, when the platform said it, and how much weight it can carry. First-party means Instagram, its documentation, or a named executive on the record.

Two rows need a note. The hashtag row is confirmed in substance, with a date we can only get secondhand: Mosseri’s wording is quoted consistently across the industry, and the nearest thing to a pin is Social Media Today’s December 2025 report, which places the remark in May 2025 and records an earlier February 2025 version of it. No first-party Instagram page carries the quote, so the May 2025 date is reported rather than read off a dated post — a weaker claim than the 22 January 2025 rows above it, and worth saying rather than smoothing over. The position itself is separately corroborated first-party, because Instagram’s own reach guidance frames hashtags as a search input and never as a reach lever.

And the per-surface rows are three years old. They are first-party, explicitly ordered, and nothing has superseded them — but any page presenting them as a 2026 statement is misleading you. Dating them is the point.

How the Feed Algorithm Ranks

Feed ranks primarily on your own past activity, then the post, then the poster, then your history with that poster. From Instagram Ranking Explained, published 31 May 2023, in Instagram’s stated order:

1. Your activity — the posts you have liked, shared, saved or commented on.
2. Information about the post — its popularity, and when it was posted.
3. Information about the person who posted — how interesting they have been to people generally.
4. Your history of interacting with that person — whether you engage with them specifically.

Instagram then predicts a set of outcomes for each candidate: how likely you are to spend time on the post, comment on it, like it, share it, or tap through to the profile. Those predictions order your feed.

One correction worth making: Feed is not a followers-only surface. It carries recommended content from accounts you do not follow, so the connected and unconnected split applies here too, not only in Explore and Reels.

How the Reels Algorithm Ranks

Reels ranking starts from the viewer’s own reel activity and their history with you, then the reel itself. Instagram’s published order, same 31 May 2023 document:

1. Your activity — which reels you liked, saved or reshared.
2. Your history of interacting with the person who posted.
3. Information about the reel — content type, popularity.
4. Information about the person who posted.

Read that ordering with Mosseri’s January 2025 statement and the two fit: the 2023 list gives the categories, the 2025 statement names the three metrics inside them that matter most. Eighteen months apart, consistent, and the newer one is the more actionable.

Reels also carries the hardest rule Instagram has published about eligibility. From Instagram’s creator guidance on improving reach: “Keep your reels short. In order to be recommended, they must be 3 minutes or less.” A binary gate, not a preference. It is dated 31 May 2023 and Instagram has changed maximum reel length before, so verify it in the app rather than trusting any article’s copy — including this one.

Because reels are recommendation-driven, most of a reel’s audience can be non-followers, which makes sends per reach the metric Mosseri’s split points at here.

How the Stories Algorithm Ranks

Stories is a relationship surface, and there is no popularity signal in it at all. Instagram’s published order:

1. Viewing history — how often you watch that account’s stories.
2. Engagement history — how often you engage with them.
3. Closeness — your overall relationship with the author.

All three are about the viewer’s existing history with you. Nothing in the published set rewards a stranger’s engagement, which is why generic reach tactics do not transfer: there is no documented mechanism for a story to break out to people who have no history with you.

The practical reading is narrow and useful. Stories performance measures your relationship with your existing audience, not your distribution. If story views fall while reel views hold, that is a retention signal about your followers, not a suppression event.

Instagram has published nothing newer on Stories ranking than the 2023 document — so if you read a 2026 article naming new Stories ranking factors, ask where they came from.

How the Explore Algorithm Ranks

Explore is the only surface where Instagram states that information about the post outranks the viewer’s own activity, and the only one where it says popularity matters significantly more. The published order:

1. Information about the post — with popularity signals weighing significantly more here than elsewhere.
2. Your activity in Explore — what you have engaged with in that surface specifically.
3. Your history of interacting with the person who posted.
4. Information about the person who posted.

The inversion makes sense once you notice what Explore is made of: almost entirely accounts the viewer does not follow. With no relationship to draw on, the post has to carry the ranking. It is also why recommendation eligibility bites hardest here — if your account is ineligible, Explore is the surface you disappear from first.

What Changed in 2026 — and What Did Not

One ranking-relevant change in 2026 could be verified first-party: on 30 April 2026, Instagram extended its unoriginal-content rule to photos and carousels. From Instagram’s announcement on rewarding original creators: “Accounts that primarily post unoriginal content in photos or carousel posts, in addition to reels, will no longer be shown in places where we recommend content.”

Three details in that announcement matter and are routinely dropped:

• It is a recommendations change, not a reach penalty on your own audience. Instagram states it “does not impact how we show people content from accounts they follow.”
• There is a stated route back, on a rolling window. Eligibility returns when most of your recently posted photos, carousels and reels are considered original within a 30-day period, calculated on a rolling basis.
• Original does not mean you uploaded the file. Instagram’s definition: “Original content is work that you wholly created or reflects your unique perspective.” Your own photography, your own design, or third-party material you edit so as to meaningfully enhance it all qualify. A border, a watermark, subtitles, or a credit in the caption do not.

What did not change matters more for reading any 2026 article. The per-surface ordering is still the May 2023 document; the three-signal statement is still January 2025. And the Views change — Views becoming the primary metric across reels, live, photos, carousels and stories — was announced by Mosseri on 7 August 2024, not in 2026, and it is a measurement simplification. Being the headline metric in Insights does not make views a ranking signal.

Clearing what Instagram recommends to you is a different mechanism with its own steps, covered in our guide to resetting recommendation feeds across YouTube, TikTok and Instagram.

Ten Instagram Algorithm Claims That Do Not Survive Checking

This section is the reason the page exists. Each of the following was found asserted as fact, usually with no attribution, across mainstream guides to this topic. None of them holds.

One pattern runs through all of them, and naming it will help you read the next article you find: the tell is precision without provenance. A range like 200–1,000, a multiplier like 3–5x, a figure like 40–60% more distribution — a number that specific should have a study behind it, and there is nothing. The claims that are supported come with a quote, an author and a date, and are usually vaguer than the folklore, because platforms speak carefully.

What Nobody Can Tell You

Publishing what is not known is part of getting this right. These questions were chased through Instagram’s blog, its creator documentation and Mosseri’s statements, and have no published answer as of 24 September 2026:

• How much more watch time matters than likes or sends. There is an ordering and the word slightly; no magnitudes, on any surface.
• Whether watch time counts replays. Widely asserted, never sourced.
• Whether Stories or Explore ranking has changed since May 2023. Nothing published since addresses either surface — the changes concern reels, recommendations eligibility and originality.
• Whether Trial Reels performance carries into ranking for the subsequent public post. Instagram describes the feature and says nothing about carry-over.
• Whether AI-generated content is downranked. No ranking statement from Meta or Instagram exists. Meta’s AI labelling is a disclosure policy, and that is not a ranking signal.
• An optimal post length, posting frequency or hashtag count. No first-party ranking source supports a number for any of them.

If a guide gives you a confident answer to one of those, that confidence is the problem.

How to Measure Yourself Against These Signals

Use the rate metrics Mosseri named, not the counts Instagram shows first. The mapping is direct, and it is the closest thing to an actionable version of the documented signals:

The reach split is the diagnostic worth running monthly. A drop in non-follower views while follower views hold is an unconnected-reach story; a drop in both is more likely content or cadence. Different causes — and averaging them hides which one you have.

The question behind most reach panics — “am I being suppressed, or did I just post something worse?” — is a separate diagnosis with its own screens and arithmetic, and we cover it in full in our guide to the Instagram shadowban and recommendation eligibility. This page is how ranking works; that one is what to do when reach falls off a cliff.

Be careful comparing engagement rate across platforms: the denominator differs, so the percentages are not comparable. Our free engagement rate calculator uses the right denominator per platform.

What This Means for How You Post

No tactic below comes with a promise, because the platform publishes no weights and anyone promising an outcome from a ranking signal is guessing. What these are: practices that follow directly from something Instagram has documented, with the signal named so you can judge them yourself.

• Keep reels under three minutes if you want them recommended. This follows from a stated eligibility gate, not a preference — the clearest rule on this page.
• Optimise for sends when you want non-followers, for likes when you want your own audience. From Mosseri’s connected/unconnected split. It changes the brief: a post built to be forwarded to one person is not the same post as one built to be liked in passing.
• Post work that is yours. From the 30 April 2026 originality rule. A border or a caption credit does not qualify, and the cost is recommendations, not your followers’ feeds.
• Treat Stories as a relationship surface. From the published Stories ordering, which contains no popularity signal at all.
• Use hashtags for search and topic clarity, not reach. From Mosseri and Instagram’s own reach guidance — while noting that Instagram’s in-app prompts have at times still claimed hashtags help you reach more people.
• Read Account Status before theorising. From the documented eligibility mechanism. Ten seconds, and it rules out the explanation people spend weeks assuming.

Doing this across platforms is where the week goes, and it is where what we build fits. Every network ranks with its own documented signals — LinkedIn’s, far thinner than Instagram’s, is covered in our companion guide to how the LinkedIn algorithm works. Outfeed AI is chat-first rather than dashboard-first: describe a post once and the assistant drafts a version per platform in your brand voice, previews each, and schedules or publishes across nine networks from one conversation. Starter is $29/month flat, Pro is $59/month flat with 3 team members, and per-seat billing starts only at the $149 per seat Agency tier. Start a free 3-day trial — 100 AI credits, card required.

Frequently Asked Questions

How does the Instagram algorithm work in 2026?

There is no single Instagram algorithm. Instagram runs a separate ranking system for Feed, Reels, Stories and Explore, and the only per-surface ordering it has ever published is from 31 May 2023. Across the recommendation surfaces, Head of Instagram Adam Mosseri said on 22 January 2025 that the three signals mattering most are watch time, likes and sends, and that creators should watch the rate versions in Insights: average watch time, likes per reach and sends per reach.

What are the top three Instagram ranking signals?

Watch time, likes and sends, per Adam Mosseri on 22 January 2025. He also split them by audience: likes matter slightly more for reaching your existing followers, sends slightly more for reaching people who do not follow you. The word “slightly” is his. Instagram has never published a weight or multiplier for any signal, so every specific figure like “sends are worth 3x a like” is invented.

Does Instagram have one algorithm or several?

Several. Instagram’s own position is that there is no single algorithm but many algorithms and ranking processes, one per surface. Stories ranks purely on the viewer’s existing relationship with the author and contains no popularity signal at all, while Explore is the only surface where information about the post outranks the viewer’s own activity. A single flat list of “Instagram ranking factors” is wrong by construction.

Do hashtags improve your reach on Instagram?

No. Mosseri has stated that hashtags are not a way to get more reach; they help people understand what a post is about and they matter for search. Instagram’s own reach guidance lists hashtags only among keyword and search surfaces, never as a distribution lever. Instagram’s in-app prompts have at times still suggested hashtags help you reach more people, which is a genuine contradiction from the platform.

Is the Instagram shadowban real?

Not as described. Mosseri has repeatedly denied that Instagram secretly suppresses reach, including to push accounts toward buying ads. What does exist is documented: content or an account can become ineligible for recommendation under Instagram’s Recommendations Guidelines, and you can check your own state in Account Status in the app. Recommendation ineligibility never changes what your existing followers see.

How long can a reel be and still be recommended?

Three minutes or less. Instagram’s creator reach guidance states it plainly: to be recommended, reels must be three minutes or less. That is one of very few binary rules Instagram has ever published about eligibility. The guidance carrying it is dated 31 May 2023 and Instagram has changed maximum reel length before, so re-check it rather than treating it as permanent.

Does posting content you did not create hurt your Instagram reach?

It removes you from recommendations, not from your followers’ feeds. Since 30 April 2026, accounts primarily posting unoriginal photos and carousels, in addition to reels, are no longer shown where Instagram recommends content. Eligibility returns once most of what you posted in a rolling 30-day window is judged original. Instagram states the change does not affect how it shows people content from accounts they follow.

Sources and Check Dates

Every ranking claim on this page traces to one of the following. All were read on 24 September 2026. Where no source exists, the text says so rather than filling the gap.

• Adam Mosseri on Instagram ranking, 22 January 2025. Source for connected vs unconnected reach, the three top signals, the “slightly” split, and the rate metrics to watch in Insights.
• Instagram — Instagram Ranking Explained, 31 May 2023. The only per-surface signal ordering Instagram has ever published, and the source for all four surface lists above.
• Instagram for Creators — Rewarding original creators on Instagram, 30 April 2026. Source for the originality rule, its extension to photos and carousels, the 30-day rolling window, the definition of original content, and the statement that it does not change what followers see.
• Instagram for Creators — Tips for improving your reach, 31 May 2023. Source for the three-minute recommendation gate, and for hashtags framed as a keyword and search input.
• Instagram for Creators — Recommendations on Instagram, 30 August 2022, updated 25 April 2023. Source for the “how many and how quickly” wording, and for recommendation ineligibility being a checkable state in Account Status.
• Adam Mosseri on Views as the primary metric, 7 August 2024, posted to his own Instagram account. Cited by date for the measurement change.

Deliberately not stated here: any numeric weight or multiplier for any signal; any test-pool size or staged-distribution ladder; any golden-hour window; any optimal post length, cadence or hashtag count. Nobody publishes those, and this page will not supply them.

Re-verification: these claims decay faster than anything else we publish, so this page is re-checked against primary sources every 90 days. The load-bearing items are the 2023 per-surface ordering, the 2023 three-minute reel gate, and anything new Mosseri says — he remains the best source on this subject.