How the Instagram Reels Algorithm Works in 2026 (What Actually Matters)
How the Instagram reels algorithm works in 2026: the signals Instagram confirms, how reels reach non-followers, and what data on 15,000 reels adds.
FypSpy research team12 min read
The Instagram reels algorithm isn't one formula. It's a set of ranking systems that predict, viewer by viewer, how likely each person is to watch a reel, like it and send it to someone. Instagram's head Adam Mosseri has said the three signals that matter most are watch time, likes per reach and sends per reach, and Instagram says every eligible reel is first shown to a small group of non-followers, then to wider audiences if it performs.
That's the official part. FypSpy tracks thousands of creator accounts every day, and our data on 15,471 reels shows the sends part in public numbers: viral reels get about 3.3× more shares per like than regular ones. Below: what Instagram has confirmed (with dates), how reels reach followers and non-followers, what our data adds, a checklist for each stage of a reel and the myths worth dropping.
How the Instagram reels algorithm works: the short version
Instagram has said it doesn't have "a singular algorithm": Feed, Stories, Explore and Reels each rank content their own way. Reels is the discovery surface; in Mosseri's 2023 explanation, the majority of what people see there comes from accounts they don't follow. A simplified model of what Instagram has described:
- Eligibility. A reel that goes against the Recommendation Guidelines (sexually suggestive content, for example), reposts someone else's video or carries another app's watermark can still reach followers but is shown less or kept out of recommendations.
- Followers. Instagram says every reel is shown to both followers and non-followers.
- A small test audience. Among non-followers, a reel is first shown to a small group of people Instagram predicts will enjoy it.
- Expansion. The best performers go to a slightly wider audience, and the process repeats. Instagram describes this as distributing content "based on its engagement, not just the creator's follower count".
- Personal ranking. For each viewer, ranking weighs their own activity, their history with the creator, information about the reel (audio, visuals, popularity) and information about the creator.
What Instagram has confirmed about ranking reels
| Signal or rule | What Instagram has said | When |
|---|---|---|
| Watch time | One of the top three ranking signals; creators should watch average watch time | January 2025 (Mosseri) |
| Sends per reach | "One of the most important signals we use in ranking": out of everyone who saw a reel, how many sent it to a friend in a DM | 2024 (Mosseri) |
| Likes per reach | One of the top three signals; likes weigh slightly more with followers, sends with non-followers | January 2025 (Mosseri) |
| Retention and the hook | Ranking uses watch time, retention, shares, likes and comments; start with a hook in the first 3 seconds | February 2025 (Instagram for Creators) |
| Predictions | How likely you are to reshare a reel, watch it to the end, like it and go to its audio page | May 2023 (Mosseri) |
| Reels shown less | Low-resolution or watermarked, muted, with borders, mostly text, or already posted on Instagram | May 2023 (Mosseri) |
| Originality | Originals replace reposts in recommendations; accounts that mostly repost lose recommendations; extended to photos and carousels with a rolling 30-day check | April 2024, April 2026 |
| Trial reels | Shown to non-followers first; can be shared with followers automatically if they perform well within 72 hours | December 2024 |
| Hashtags | Don't increase reach; up to five per post | 2025 |
Primary sources: Instagram's ranking explainer, its February 2025 post on how reels are ranked and recommended and the April 2026 originality update.
Watch time and retention
Watch time tells Instagram whether people actually stay. Instagram's February 2025 advice to creators was to "start with a compelling hook in the first 3 seconds" and to track how many viewers keep watching past that point. Reels Insights now show skip rate (the share of viewers who swipe away in the first three seconds) and a retention graph, so you can see exactly where people leave.
Sends per reach
Mosseri has called sends per reach one of the most important ranking signals and described it simply: out of all the people who saw your reel, how many sent it to a friend in a DM. In January 2025 he added a nuance: likes are slightly more important for content from accounts people follow, sends for content from accounts they don't. Sends are the signal for reaching new people. He also told creators not to force it; the reel has to be worth sending.
Likes per reach
Likes still count, mostly with existing followers. That's why a reel can have a lower like rate than usual and still travel far: once it reaches strangers, likes per view fall, and what matters is whether those strangers keep watching and send it on.
Originality
Since 2024, when Instagram finds the same content posted more than once, it recommends the original instead of the repost, and accounts that mostly post content they didn't create or materially edit lose recommendations. On April 30, 2026 Instagram extended the rule to photos and carousels; an account regains eligibility when a significant share of its recent posts over a rolling 30 days are original. Adding a border, a watermark, subtitles or a credit doesn't make someone else's video yours. To build on a proven idea the right way, see how to recreate a viral reel.
How reels reach followers and non-followers
Reels reach two audiences at once. Followers see them in Feed and the Reels tab; non-followers mostly find them through recommendations, which start with the small-audience test described above, and through DMs from friends. Insights show the split between followers and non-followers for every reel.
Because distribution depends on engagement and not only on follower count, small accounts often reach well beyond their followers. Typical views (the median of an account's recent reels) by account size in FypSpy's data:
| Followers | Typical views per reel | As a share of followers |
|---|---|---|
| Under 10K | ~5.5K | ≈109% |
| 10–50K | ~11.3K | ≈49% |
| 50–100K | ~19.4K | ≈28% |
| 100–500K | ~37K | ≈18% |
| 500K–1M | ~86K | ≈11% |
| 1M+ | ~173K | ≈11% |
A typical reel from an account under 10K followers is seen more times than the account has followers; at 1M+ typical views equal about a tenth of the follower count. Views grow with audience size, but much more slowly than followers. Check any account with our reel views calculator, which also gives the ×N of a single reel.
Eligibility also works at the account level. If an account repeatedly posts content that goes against the Recommendation Guidelines, Instagram says the whole account may become ineligible for recommendations for a period, and none of its content will be recommended to non-followers. Account Status in the app shows whether that's happening.
What FypSpy's data adds (November 2026)
FypSpy scans thousands of Instagram creator accounts every day and records Instagram's own numbers for each reel. For rates we use 15,471 reels (from 3,512 accounts) that were older than 72 hours, so their numbers had settled. "Viral" means 5× or more the author's typical views.
Shares are the only interaction that rises on viral reels
| Median per reel | Regular reels | Viral reels (5×+) |
|---|---|---|
| Likes per 1,000 views | 30 | 14 |
| Comments per 1,000 views | 1.2 | 0.4 |
| Saves per 1,000 views | 2.6 | 1.6 |
| Shares (sends) per 1,000 views | 0.5 | 0.8 |
| Shares per 100 likes | 1.8 | 6.2 |
Viral reels travel to strangers, who like, comment and save less often than fans do, so those rates fall. Shares are the only interaction that rises per view, and relative to likes they're about 3.3× higher. That's sends per reach showing up in public numbers. Our rates are per view, not per reach (views include replays), so treat them as a close proxy rather than Instagram's internal metric.
The practical reading: a low like rate on a reel with big reach is normal; a low share rate is the real warning sign. Each metric is explained in Instagram reels metrics explained.
Most reels stop at the first rings
Among reels older than 72 hours, 15.9% reached 2× their author's typical views, 5.9% went viral at 5× and 2.6% reached 10×. Only one reel in ten got 3× or more, and one in a hundred got about 19× or more. The median viral reel ended up around 158K views. That's the small-audience test at work: most reels stop early, and a few keep expanding. More cuts are on the Instagram reels benchmarks page.
Length, audio, format and timing
| Factor | What we found | What to do |
|---|---|---|
| Length | Under 8 s: 4.5% went viral; 8–15 s: 6.6%; 15–30 s: 6.3%; 30–60 s: 6.5%; 60 s+: 5.9% (fewer reels in the longer brackets) | Keep most reels between 8 and 60 seconds |
| Audio | Original audio 6.2% vs licensed music 5.3% | Your own voice or original sound by default |
| Format | POV skit 7.0%, vlog 6.9%, transition 6.9%, dance 6.3%, talking to camera 6.0%, text on screen 6.0%, lip-sync 5.2%, try-on 5.0%, cinematic 4.5% | Story and personality beat pure aesthetics |
| Posting time (New York) | Best 3-hour windows 06–09 (6.8%) and 12–15 (6.6%); weakest about 5.1% | A small lever; consistency matters more |
| Cadence | The median account posts about 6 reels a week; 18% of accounts had at least one 5× reel in a week | Post steadily and keep testing |
To see what gets sent in your niche, look at outliers sorted by shares: FypSpy shows every reel's views, likes, comments, shares and saves, and sorts outliers by views, likes, comments or shares.
A checklist for every stage of a reel
Every signal Instagram names maps to a stage of the reel. Work through them in order: hook → retention → share trigger → rewatch.
1. Hook: the first 3 seconds
- Clear in half a second: who it's for and what's happening.
- On-screen text visible from the first frame and short enough to read at a glance.
- Understandable to strangers. Instagram's own advice: "If your goal is to reach new people, your reels should make sense to those who may not know you."
- A high skip rate in Insights means the opening failed, not necessarily the idea.
- Proven openings are collected in Instagram reel hooks.
2. Retention: the middle
- One idea per reel; cut the slow start, the pauses and the last dead second.
- Deliver the payoff before attention drops; the retention graph shows where it does.
- Keep the sound on: Instagram has said muted reels are shown less.
3. Share trigger: a reason to send it
- Relatable ("this is you"), useful, funny or debatable: something a viewer wants to show one specific friend.
- Don't beg for shares. Mosseri's advice was not to force it, and Instagram lists engagement bait among content it tries not to recommend.
- Ask before posting: who exactly would send this, and to whom?
4. Rewatch: a reason to watch again
- A seamless loop, where the end runs back into the start.
- A detail people only catch on the second viewing, or text that takes a second read.
- An ending that recasts the beginning.
5. After posting
- Reply to early comments, especially from new viewers.
- Judge the reel after 72 hours against the account's typical views, not against other accounts.
- Test new hooks as trial reels, which go to non-followers first, and turn winners into a series.
Instagram algorithm myths and what's actually true
| Myth | What's true |
|---|---|
| Hashtags get reels onto Explore | Mosseri has said hashtags don't increase reach, and since December 2025 Instagram has been limiting posts to five |
| There's a magic time to post | In our data the best and worst 3-hour windows differ by less than two percentage points (6.8% vs about 5.1%); see best time to post reels |
| Big accounts get pushed more | Instagram says distribution is based on engagement, not just follower count; small accounts often out-reach their follower count |
| Shorter is always better | Reels under 8 seconds went viral least often in our data |
| Likes are what matter | For reaching non-followers, sends matter more than likes |
| Reposting a viral reel is a shortcut | Instagram replaces reposts with originals and stops recommending accounts that mostly repost |
| You can reset your account's algorithm | Instagram hasn't described any reset for creators; each reel is ranked on how viewers respond, and account-level limits show up in Account Status |
| Every drop is a shadowban | Most drops are normal variance; check Account Status first, then the reel itself with why reels aren't getting views |
FAQ
Did the Instagram reels algorithm change in 2026?
The core signals Instagram has described publicly (watch time, likes per reach and sends per reach) come from its 2025 explanations, and the small-audience-first testing from its 2024–2025 posts. The clearest 2026 change is originality: on April 30, 2026 Instagram extended its rules against unoriginal reposts to photos and carousels, with eligibility rechecked over a rolling 30 days. Late 2025 also brought the five-hashtag limit and "Your Algorithm" topic controls for viewers.
How do I reset my Instagram reels algorithm?
As a viewer, you can. Meta announced a "Reset suggested content" option in November 2024 (profile → menu → Content preferences → Reset suggested content) that clears recommendations in Explore, Reels and Feed, and since December 2025 Instagram has been rolling out "Your Algorithm" in the Reels tab, which shows the topics behind your recommendations and lets you edit them. As a creator there's no reset button: each new reel is ranked on how viewers respond, and Account Status shows any account-level limits.
Is the Instagram reels algorithm broken?
Usually not; reach is simply lopsided. In our data only 5.9% of reels reach 5× their author's typical views, and roughly half land below typical by definition, so big swings between reels are normal. If every reel drops at once, check Account Status for recommendation limits, then compare recent reels with your typical views before changing strategy.
How do reels reach non-followers?
Instagram says each eligible reel is first shown to a small group of non-followers it predicts will enjoy it, and the best performers go to progressively wider audiences. Sends weigh slightly more than likes with these viewers. Small accounts benefit most: in our data, accounts under 10K followers typically get views equal to about 109% of their follower count.
What matters most for the reels algorithm?
According to Mosseri: watch time, sends per reach and likes per reach, plus eligibility, meaning original content that follows the Recommendation Guidelines. Our data adds that viral reels earn about 3.3× more shares per like than regular ones, so build every reel around a reason to send it.
Do hashtags help reels get more views?
Not really. Mosseri has said hashtags don't increase reach and are mainly a way to tell people what a post is about. Since December 2025 Instagram has been limiting posts and reels to five hashtags, so use a few specific ones and put the effort into the hook.