How Instagram Algorithm Works in 2026: What Actually Determines Reach?

A client sent me two Reels last spring and asked me to explain the difference.

The first one took her four days. Proper lighting, three locations, a script she rewrote twice, colour grading, custom captions. It went out on a Tuesday morning and did about 900 views. Nine hundred, on an account with 41,000 followers.

The second one she filmed on her phone in a car park while waiting for a delivery. Bad audio. One take. She almost didn't post it. It did 640,000 views in five days and brought in roughly 3,000 followers.

Her question was the obvious one: what did I do wrong on the first one?

She hadn't done anything wrong, exactly. She'd made the assumption almost everyone makes — that Instagram rewards effort, or quality, or how long something took. It doesn't. It rewards predicted behaviour: what a specific person will do in the next few seconds. Keep watching, scroll away, tap the profile, send it to a friend. The four-day Reel was beautiful and slow. The car park Reel had a first line blunt enough that people stopped mid-scroll and then sent it to someone who needed to hear it.

I've been doing SEO and social growth for a bit over eight years, and Instagram's ranking systems are easily the most misunderstood thing I work with. Not because they're secret — Instagram has published more about ranking than most platforms ever have — but because the explanations that spread fastest are the ones that sound like hacks. Post at 7pm. Use 30 hashtags. Reply to every comment within an hour.

So here's the real version: how each surface ranks content in 2026, which signals genuinely matter, which ones people waste months on, and what I'd change first if you handed me your account tomorrow. Nobody outside Meta sees the actual weights, so where something is uncertain I'll say so rather than pretend.

What Is the Instagram Algorithm?

Let's kill the biggest misconception first.

There is no such thing as "the Instagram algorithm," and there hasn't been for years. Instagram runs separate ranking systems for each place content appears — Feed, Reels, Stories, Explore, Search. They share underlying machinery but weigh signals very differently, because people use those surfaces for completely different reasons.

That isn't pedantry. It's the single most useful thing you can internalise. When someone says "the algorithm buried my post," the first question is which one? A carousel that flops in Feed and a Reel that flops in Explore have failed for different reasons and need different fixes.

How ranking actually happens

Every one of those systems works in roughly the same three stages:

1. Candidate gathering. For a given user in a given moment, Instagram assembles an enormous pool of content it could show — posts from accounts they follow, Reels from creators they don't, Explore candidates from topics they've engaged with.

2. Prediction. Models estimate the likelihood of specific actions for each candidate. How long will this person watch? Will they like it, comment, send it on — or hide it?

3. Ranking. Those predictions combine into a score, and the highest-scoring content goes first.

That's the whole thing. Everything else — hooks, captions, saves, watch time — is just an input into those predictions.

Connected vs unconnected reach

The other framing that matters enormously: Instagram distributes content on two separate tracks.

Connected reach is people who already follow you. Even here, not everyone sees everything. Your post competes against every other post that person could see, and Instagram ranks based on how relevant it predicts your content is to them specifically.

Unconnected reach is people who don't follow you. This is recommendations — Reels feed, Explore, suggested posts in the home feed. This is where growth comes from, and the bar is significantly higher, because Instagram is risking a stranger's attention on you.

Almost every "why is my reach down" conversation I have turns out to be a collapse in one of these two, not both. Pull up your Insights and split reach by followers vs non-followers before you diagnose anything. I've seen accounts panic about a shadowban when their follower reach was perfectly healthy and their non-follower reach had simply stopped, which is a content problem, not a penalty.

The three signals Mosseri keeps naming

Across surfaces, Instagram's leadership has been unusually consistent about which signals carry the most weight:

• Watch time — how long people actually spend with your content

• Sends per reach — how often people share it via DM, relative to how many saw it

• Likes per reach — the like rate, relative to reach

Sends per reach is the one most people still underweight. A like costs nothing. Sending something to a specific person costs a little social capital — you're saying "this is worth your time, and I thought of you." Instagram reads that as a much stronger quality signal, particularly for reaching non-followers.

"Your Algorithm" and user-side control

Something genuinely new since late 2025: users can see and edit the topics Instagram thinks they're interested in. The "Your Algorithm" controls launched in December 2025 and rolled out broadly through 2026, letting people add topics, remove ones they don't want, and reset their recommendation history entirely.

For creators, the implication is subtle but real: topic clarity matters more than it used to. If someone has explicitly told Instagram "more film photography, less fitness," your content has to be classifiable enough to land in the right bucket. Accounts posting about eleven unrelated things are harder to categorise and harder to deliberately opt into. I've watched this with niche pivots — the first month is usually rough, because the system is still routing you to the old audience. That's lag, not punishment.

How Instagram Feed Algorithm Works

Feed is where your existing audience lives. It's also, increasingly, a recommendation surface — Instagram now mixes in suggested posts from accounts you don't follow, so Feed isn't purely a follower channel anymore.

For ranking, Feed leans harder on relationship than any other surface except Stories.

Relationship and interaction history

Instagram builds an implicit model of how close each viewer is to each account. It's assembled from things like:

• How often that person has liked, commented on, or saved your posts before

• Whether they DM you or you DM them

• Whether they've searched for your handle

• How often they visit your profile

• Whether they watch your Stories

• Whether you're mutually connected or tagged in each other's content

Practically, this means your reach among followers isn't one number. It's thousands of individual relationship scores. When reach drops, it's often because a chunk of your audience quietly stopped interacting months ago and the system slowly demoted you for them.

I had a food brand client whose Feed reach kept sliding even as their content got better. The cause was a giveaway campaign a year earlier: thousands of followers who'd never engaged with anything since. The account looked bigger and performed smaller.

Interest prediction

Beyond relationship, Feed predicts topical interest. Instagram reads your content through the frames themselves, the caption, on-screen text, audio, location, tagged accounts, and how similar users responded to similar material. Which is why consistency in subject matter helps — not consistency of format, since you can mix Reels and carousels freely, but consistency in what your content is about.

Recency

Feed isn't chronological, but recency is still a real signal, balanced against relevance. In practice your post has a window — usually the first few hours — where it competes well on freshness. If early performance is weak, it doesn't get to keep coasting on that.

Session behaviour

Nobody thinks about this one. Instagram adapts within a session: how long someone's been in the app, whether they've already seen four Reels in a row, how fast they're scrolling, whether they've hit your content already today. It's part of why the same post performs differently depending on where in a session it surfaces — and why posting five times in a day cannibalises rather than multiplies reach.

In Feed specifically, the signals that move the needle most are time spent on the post, sends, saves, substantive comments, profile visits, and follows — roughly in that order. Likes still count, but they sit near the bottom. Negative feedback sits near the top, in the wrong direction. There's a full breakdown further down.

Carousels deserve a mention. They're one of the few formats that raise time-on-post without video, because the viewer has to swipe. Instagram also re-surfaces carousels to people who didn't engage the first time, starting from a later slide — I've had carousels quietly gather reach for a week after posting, which almost never happens with a single photo.

How Instagram Reels Algorithm Works

Reels is the growth engine. It's the surface most likely to put you in front of strangers, and it's ranked almost entirely on behavioural signals rather than relationship.

Watch time is the foundation

Everything starts with retention. Not views — views just count who got served the thing. What matters is what happened after the first frame:

• Average watch time — how many seconds people actually stay

• Completion rate — what proportion watch to the end

• Replays — the strongest version of the same signal, since a loop means someone chose to watch twice

• Early drop-off — how many bail in the first two or three seconds

That first two-to-three seconds is genuinely disproportionate. Most viewers are scrolling with sound off, so whatever justifies stopping has to be visible. Not clever. Visible. Text on screen, motion, a face mid-sentence, something visually odd.

Length matters, but not the way people assume. Short Reels hit high completion rates more easily, which is why very short clips look good on paper. But a longer Reel that holds most of its viewers generates far more total watch time than a nine-second one that holds nearly all of them. If you can hold attention, longer is better. If you can't, don't stretch it.

Sends and shares

This is where Reels growth actually comes from. If watch time gets you distribution, sends get you expansion — a DM share tells Instagram the content is worth injecting into a private conversation, which is about the highest compliment the system can register.

So the useful question before posting isn't "is this good?" It's "who would someone send this to, and what would they type?" If you can't answer that, the Reel will probably plateau at your existing audience. It's also why paid support has such a hard ceiling here: buying Instagram Reel views can warm up a number, but it will never manufacture a send — and sends are what the recommendation system is actually shopping for.

Originality, reposts, and watermarks

This got significantly stricter in 2026. In April, Instagram extended its originality rules from Reels to photos and carousels: accounts that mainly post content they didn't create are no longer recommended to people who don't already follow them.

The details are worth knowing exactly:

• Content is original if you created it or materially transformed it

• Watermarks, speed changes, borders, or screenshots of someone else's post with the handle visible do not count as transformation

• Memes still work — but the creator has to add the joke, commentary or perspective that wasn't there

• Followers still see aggregator content normally; it's recommendations that get cut

Separately, visible watermarks from other platforms have been a distribution drag for a long time. Mosseri has been clear that Instagram doesn't penalise videos edited in other apps — it penalises the logo. Export clean.

I've seen this bite good accounts. A travel page I worked with was recycling TikToks with the watermark cropped but a ghost of it still visible along the edge. Non-follower reach sat at basically zero for months. We re-exported everything from the source files and Explore distribution came back within about three weeks. One case doesn't prove causation — but I've watched that same story play out enough times to take it seriously.

Audio

Audio does two jobs: it's a discovery pathway, since people browse by sound, and it's a classification signal that helps Instagram place your content in context.

But trending audio is dramatically overrated as a growth lever. It helps at the margin. It does not rescue a weak first three seconds — I've tested identical Reels with trending versus original audio and the difference was noise compared to what a better hook did. Original audio has its own advantage: if other people use it, that's attribution pointing back at you.

Creator history and topic matching

Instagram maintains a sense of what your account is about and how it has performed with different audience clusters. New accounts have it harder — not because of a "new account penalty," but because there's no history to predict from, so the system tests cautiously.

Which is why Trial Reels are useful: publish to non-followers only, get 24 hours of cold-audience data, then decide whether your actual audience sees it. Mosseri has been upfront that trials almost always get less reach than normal Reels, so the only fair comparison is trial against trial. It's the closest thing Instagram has to a test environment.

How Stories Are Ranked

Stories are a different game, and most brands treat them as a broadcast channel when they're really a relationship channel.

Every Story comes from an account you follow, so there's no discovery layer to optimise. The only question is tray position — near the front, or buried past the point where anyone keeps tapping?

Instagram has described the main Stories signals as:

• Viewing history — how often this person watches your Stories. The strongest one.

• Engagement history — replies, reactions, sticker interactions, DMs sparked by your Story

• Closeness — the overall relationship model, including DMs and mutual connections

• Recency — more recent Stories rank ahead of older ones, though a close connection's older Story will usually still beat a distant account's fresh one

Then there's how people responded to this Story: completion, skip rate, exits out of the viewer entirely, and profile taps. Skip rate is the metric almost nobody checks and everybody should. If frames three through seven get hammered with forward-taps, that's a direct instruction from your audience.

What actually improves Stories performance

• Post consistently, not constantly. Daily presence beats twelve frames twice a week; tray position decays when you go quiet.

• Front-load. Frame one decides whether anyone sees frame six.

• Ask for replies, not just taps. A poll is fine, but a question that earns a real DM reply is worth far more, because DM interaction feeds the closeness model.

• Use Close Friends deliberately. Completion tends to be high there, because the audience self-selected.

• Cut the filler. Eight frames where three would do just trains people to skip you.

One pattern I've seen repeatedly with brand accounts: Stories performance improves when they stop posting about the brand and start posting from it. Behind the scenes, a real question, a mistake, a bad day. Relationship signals respond to human things, and Stories is where that shows up first — which is why I read Story completion, rather than raw Instagram Story views, as my first check on whether an audience is still warm.

How Explore Works

Explore is pure recommendation. Everything there comes from accounts you don't follow, selected because Instagram predicts you'll like it — which makes it the highest bar of any surface, since there's no relationship credit to draw on. Roughly, the pipeline works like this:

Interest clusters. Instagram groups users and content into topic and behaviour clusters. Engage with three ceramics accounts and you land in a neighbourhood with people who do the same — so content that performed well with them becomes a candidate for you.

Content similarity. The system reads your post from its visuals, text, audio and metadata, then finds audiences that responded to similar material.

Engagement velocity. How fast a post earned meaningful engagement relative to how many saw it. Explore is more sensitive to early performance than Feed.

Seed signals. What you've saved, lingered on, and visited seeds what Explore shows you next.

For creators, three things follow from this.

First, first impressions are everything. An Explore viewer has no reason to be generous. If your content needs context to land, it won't land.

Second, niche beats broad. Explore matches clusters, so a tightly-defined account is easier to route. I've seen non-follower reach climb sharply on accounts that did nothing but narrow their topic — counterintuitive until you understand the mechanism.

Third, new creators do get surfaced. Explore isn't a closed club — Instagram constantly tests small accounts because it needs fresh supply. Small accounts don't get less reach for being small; they get less because there's less data behind them and the content is usually less refined.

Instagram Ranking Signals Explained

Here's the consolidated picture. Weights are my read based on Instagram's public statements and observed behaviour across accounts — nobody outside Meta has the actual numbers, so treat this as a well-informed ordering, not gospel.

On negative signals

People underestimate these. One "not interested" tap counts for more, in the wrong direction, than several likes do in the right one. That's why engagement bait and misleading hooks are so expensive: they buy a moment of attention and pay for it in irritation, and irritation is a measured signal. A caption that promises what the content doesn't deliver isn't gaming the system — it's feeding it evidence against you.

Biggest Instagram Algorithm Myths

I could write this section entirely from client Slack messages. Let's go through the ones that come up most.

"I've been shadowbanned." Usually no. Instagram doesn't use the term, and Mosseri has said directly that connected ranking doesn't limit reach. What does exist is narrower and documented: content breaking the Recommendation Guidelines becomes ineligible for recommendations, and spam-like behaviour can restrict distribution. Both show in Settings → Account Status, the only source of truth here. Check it before you spend a week theorising. Nine times out of ten, non-follower reach simply dried up because the content stopped earning it.

"Hashtags drive reach." Not anymore, and this one is now settled by platform rule rather than opinion. Instagram removed the ability to follow hashtags in December 2024, then in December 2025 announced a cap of five hashtags per post and Reel, rolling out gradually. Their stated reasoning was that fewer, more targeted tags work better than piles of generic ones. Hashtags today are classification metadata — a hint about what your content is about — not a distribution lever. Use two to five relevant ones and move on.

"There's a magic posting time." There's a good time for your audience, visible in Insights, and it's worth using. But it's a tiebreaker, not a strategy. I've A/B'd posting times for years: the gap between a good hour and a mediocre one is real but small. The gap between a good hook and a weak one is enormous.

"Business accounts get less reach." This one refuses to die. There's no evidence for it, and structurally it makes no sense — Instagram has no reason to suppress the accounts most likely to buy ads. What's true is that business accounts tend to post more promotional content, and promotional content performs worse. Correlation, not mechanism.

"External links kill reach." Also no. There's no meaningful evidence of a link penalty in captions or Stories. What happens is that link-driven posts tend to open with a call to action instead of a hook, and people leave the app — which does mean less time on the post.

"Editing a caption tanks the post." No. Edit typos freely — I've tested this repeatedly and found nothing. The myth probably survives because people edit posts that were already underperforming, then blame the edit.

"Delete a bad post so it doesn't drag your account down." There's no cumulative penalty score that a bad post adds to. Deleting removes whatever residual engagement it was still earning, and if it was gaining any reach at all you've just thrown it away. Archive if you dislike it visually. Don't delete out of superstition.

"Engagement pods help." They help the number and hurt the signal. Pods generate engagement from people with no real affinity for your content, which pollutes the interest model Instagram is building about your audience. I've inherited accounts that ran pods, and the pattern is consistent: healthy-looking engagement rate, terrible non-follower reach.

"Replying to comments in the first hour boosts the post." Replying is good for community and does add comment volume. But there's no evidence of a golden-hour multiplier. Reply because it's worth doing, not because a timer is running.

Instagram SEO in 2026

This is the part of Instagram growth that's changed most in the last two years, and it's still underexploited by almost everyone I work with.

Instagram search stopped being handle-matching a while ago. It now reads captions, on-screen text, audio, alt text, location and the visuals themselves — which makes Instagram functionally a search engine, and one where the competition is far softer than Google.

Captions as searchable text

Write captions the way you'd write a page title and opening paragraph: say what the thing actually is, in the words a real person would type. "Obsessed with this ✨" tells Instagram nothing. "Three-ingredient Greek yoghurt flatbread you can make in a pan" tells it the dish, the method and the constraint. The first line matters most, for search and because it's what shows before the "more" cutoff. And you don't need to keyword-stuff — one clear description beats five awkward repetitions.

The name field

The most neglected free win on Instagram. Your username is your handle; your name field — the bold line under it — is separately indexed and searchable.

If you're a Lisbon tattoo artist and your name field says "Marco," you're invisible. If it says "Marco | Fine Line Tattoo Lisbon," you show up when someone searches that. I've made this single change on client accounts and watched search-driven profile visits climb within a couple of weeks. It takes thirty seconds.

Alt text

Instagram auto-generates alt text, but you can override it under advanced settings when posting, or afterwards via edit. It serves accessibility, it feeds Instagram's understanding of the image, and since Instagram content became eligible for Google indexing, external search engines can read it too. Write it descriptively and honestly — not "best coffee shop Lisbon coffee Lisbon cafe." Describe what's actually in the frame.

Profile SEO

• Bio: what you do and who it's for, in plain words. Nobody searches "living my truth."

• Category: pick an accurate one on a professional account.

• Pinned posts: three slots that define first impression. Use your most representative work.

• Highlights: name them with real words, not symbols.

• Location: tag it if you're local — still a genuine discovery pathway for local businesses.

The Google indexing change

Since July 10, 2025, public content from adult-owned professional accounts has been eligible for indexing by Google and other search engines — photos, videos, carousels and Reels posted from 2020 onward. It's on by default and can be switched off under Settings → Privacy.

This matters more than it sounds. Instagram posts historically had a shelf life of about 48 hours; an indexed post can keep pulling visitors months later. My honest take after watching it for a year is that it hasn't transformed anyone's traffic overnight — but for accounts producing specific, searchable content it's a slow compounding gain that costs nothing extra. Write captions as though someone might find them on Google, because they might.

How to Increase Reach Under Today's Algorithm

Prioritised. Do the top ones first — the ordering matters more than the list.

1. Fix your first three seconds before anything else. Nothing else here compensates for a weak opening. Watch your own Reels with the sound off and ask honestly whether you'd stop. If it opens with a logo animation or "hey guys, so today," you're donating your reach to whoever's next in the feed.

2. Make things worth sending. Before posting, name the person someone would send it to and what they'd type. Sends come from content that's genuinely useful, sharply relatable, surprising, or settles an argument. They don't come from generic inspiration, promos, or anything that only makes sense if you already follow the account.

3. Narrow your topic. Pick a lane the recommendation system can route, and widen later once you have authority in a cluster. This is the most common structural fix I make and the one people resist hardest.

4. Post original, clean exports. No other platform's watermark. No reposts without genuine transformation. If you're recycling TikToks, export from the source file.

5. Be consistent rather than heroic. Three or four solid posts a week sustained for six months beats two weeks of daily posting followed by a burnout gap.

6. Build in a reason to save. Checklists, recipes, step-by-steps, "screenshot this" moments. Saves are among the strongest signals available and the easiest to design for deliberately.

7. Use Stories for relationships, not broadcasting. Daily, short, with at least one thing that invites a reply. It feeds the relationship model that drives Feed ranking.

8. Test with Trial Reels. Especially for new formats or a niche pivot. Compare trials to trials, never to your normal Reels.

9. Treat captions and your name field as search assets. Ten minutes of profile SEO is the cheapest reach you'll ever buy.

10. Read your Insights properly. Split follower vs non-follower reach. Read retention graphs, not view counts. Look at what your most-sent posts have in common. Most accounts are sitting on the answer already.

11. Collaborate. Collab posts appear on both profiles and draw from both audiences. Still one of the highest-leverage organic moves available, and badly underused outside large creators.

If you want a deeper walk through the tactical side of that list, I go further into content structure and hook writing in my guide to growing faster on Instagram.

Common Mistakes That Kill Reach

These are the ones I see over and over, in roughly the order of how often they come up.

Chasing followers instead of engagement. I've reviewed accounts with six figures of followers and reach in the low thousands. Follower count is a stock; engagement is a flow, and only the flow drives distribution. It's also why cheap growth backfires so specifically — an audience that never engages permanently dilutes your relationship scores. If you do choose to grow Instagram followers with paid support, where they come from matters vastly more than how many.

Engagement bait. "Comment YES if you agree." It generates comments and it generates hides, and the hides cost more.

Ignoring negative signals. People obsess over likes and never check where viewers dropped off or which posts drove unfollows. The negative data is more diagnostic.

Deleting and reposting. I've seen creators kill a Reel two hours in because it "wasn't performing." Two hours is nothing. I've watched Reels sit flat for four days and then take off — recommendation surfaces don't run on Feed's clock. Give it a week.

Overposting. More posts in a day usually means each one gets less. Your content competes against itself for the same viewers in the same sessions.

Treating a viral moment as the new baseline. Reach almost always regresses after a big Reel. You picked up a wave of loosely-interested followers, and the next post gets measured against a larger, colder audience. That's arithmetic, not a penalty.

Judging health by follower count alone. A number that only goes up tells you very little, and it updates more slowly and less precisely than people assume. A dedicated live follower counter beats refreshing the app if you want to watch an account move, but it still won't tell you whether anyone is actually watching.

Frequently Asked Questions

Does Instagram have one algorithm or several? Several. Feed, Reels, Stories, Explore and Search each run separate ranking systems with different signal weights. They share underlying infrastructure but behave differently, which is why content can succeed on one surface and fail on another.

What is the single most important ranking signal in 2026? There isn't one universal answer, but the two that consistently matter most are watch time and sends per reach. Watch time gets you distribution; sends get you distribution to people who don't follow you.

Why did my reach suddenly drop? Split reach by followers vs non-followers first. If follower reach is stable and non-follower reach collapsed, it's a content or recommendation-eligibility problem. If both dropped, look at posting consistency, topic changes, and recent negative feedback — then check Account Status. Reach also fluctuates naturally; not every dip has a cause you can act on.

Do hashtags still work? They classify, they don't amplify. Instagram capped posts and Reels at five hashtags starting December 2025, and removed hashtag following back in December 2024. Use a few genuinely relevant ones. Don't build a strategy on them.

Is the shadowban real? Not as commonly described. What's real: content violating the Recommendation Guidelines becomes ineligible for recommendations, and spam-like behaviour can restrict distribution. Both show in Settings → Account Status. If nothing's flagged there, your reach problem is a content problem.

Does posting time matter? A little. Post when your audience is active — Insights will tell you. Don't spend more than five minutes on it.

How often should I post? Whatever you can sustain for six months. A few Feed posts and a couple of Reels a week with daily Stories works well for most accounts. Volume without consistency is worse than consistency without volume.

Do Reels get more reach than photos? Reels reach non-followers more easily, because that's the surface built for recommendations. But each surface ranks separately, and carousels frequently beat Reels inside Feed, especially on saves and shares. Instagram doesn't "prefer" Reels — Reels is just where the discovery traffic lives.

Are saves more important than likes? Yes, meaningfully. A save signals intent to return; a like is a reflex. Both count, but saves carry more weight and are easier to earn deliberately.

Will reposting content from TikTok hurt me? If it carries another platform's watermark, yes — visible watermarks reduce distribution. And under the originality rules extended in April 2026, accounts that mainly post content they didn't create or meaningfully transform aren't recommended to non-followers. Reposting your own content, exported clean from the source file, is fine.

Can I reset the algorithm? As a viewer, yes — Instagram's "Your Algorithm" controls let you edit your interest topics and reset recommendation history entirely. As a creator, there's no reset button for how Instagram classifies your account. The only path is consistent posting in a clear lane until the model updates.

Does buying engagement help distribution? It depends entirely on the source, and it's rarely the lever people think it is. Engagement from accounts with no real affinity for your niche doesn't produce the behavioural signals — watch time, sends, saves — that recommendation systems weigh, and poor sources dilute the audience model Instagram builds about you. If you use paid support at all, treat it as a warm-up rather than a substitute, and be selective about where it comes from. Our Instagram growth services page explains how we approach sourcing.

Final Thoughts

Here's the thing I wish someone had told me six years ago, before I spent a lot of hours reverse-engineering things that turned out not to matter.

You cannot beat the algorithm, because the algorithm isn't your opponent. It's a very expensive machine for predicting whether a human being will find your content worth their time. Every "hack" that ever worked, worked because it happened to align with that. Every one that stopped working, stopped because it drifted away from it.

Which means the durable strategy is embarrassingly unglamorous: make things specific people actually want to watch, in a lane clear enough for the system to route, consistently enough for it to learn who those people are.

The accounts I've watched grow most reliably weren't the ones with the best grasp of ranking signals. They were the ones that showed up every week for a year, noticed which posts got sent to friends, and made more of those. The ones that stalled either changed direction every few weeks or quit at month four because month three was slow.

Month three is always slow. That's not the algorithm. That's just what compounding feels like from the inside.

Learn the mechanics — they'll stop you wasting time on hashtag counts and posting-time superstition. Then put them down and go make something worth sending to a friend.

Instagram's ranking systems change continuously and Meta doesn't publish the weights. Everything here reflects Instagram's public documentation, Adam Mosseri's statements, and patterns observed across accounts as of August 2026. If something here stops matching what you see in your own Insights, trust your Insights.