How the Instagram Algorithm Works (What Nobody Tells You)

Why Most Instagram Algorithm Advice Stops Too Early

Most articles on how the Instagram algorithm works will hand you a list of signals. Watch time. Likes. Shares. Saves. That is not wrong. Those signals are real.

But they stop there. They tell you the levers exist. They do not tell you why the levers work for some creators and not for others.

The missing piece is this: the algorithm does not care how strong your hook is if the topic requires knowledge your viewers do not have yet. And if you are reading this, there is a very good chance the majority of your content has that problem right now.

Shut up and give me the value. Fine. Here it is. Let’s crack into it.

How Does the Instagram Algorithm Work?

The algorithm’s job is to find videos people will watch all the way through and then share. It measures watch time, likes per reach, and DM shares above everything else. DM shares are the most powerful of the three for reaching people who have never heard of you – and that fact is the one almost nobody is teaching.

Before we get to the signals, you need to understand how the algorithm tests your content before it reaches anyone you already know.

Why Is My Video Stuck Under 1,000 Views?

Every public Reel goes through an audition. Instagram shows it to roughly 200 to 1,000 non-followers first and watches what happens. If those cold accounts watch to completion, replay it, or DM it to a friend, distribution expands. If they scroll away in two seconds, it stops. The first 90 minutes after you post is when this happens.

Your existing followers often see the video after it has already passed or failed.

Hold up. The people deciding whether my video goes anywhere are people who have never heard of me?

Exactly. And this is the part that changes how you should think about every content decision you make.

When a video stalls, most creators assume the hook was weak. They study hook formulas. They rewrite the opening. They test new formats. The video still stalls.

That is because the problem is upstream of the hook.

When a stranger hits your video in the audition pool, they decide within seconds whether this topic is relevant to them at all. If the topic only makes sense to someone who already knows your terminology and follows your work closely, they scroll. Every time. No hook rewrite fixes a topic that requires prior knowledge to find interesting.

What Is the CCN Framework?

CCN stands for Core, Casual, and New. These are the three audience tiers every piece of content you post is aimed at, whether you realize it or not. Core content serves your committed followers. Casual content reaches your broader niche. New-audience content is the only tier that consistently passes the cold audition and grows your account with strangers.

Think of three circles.

The innermost circle is your core audience. These are people who follow you closely. They know your language. They track what you teach. If you post something highly specific to your method or framework, they love it.

But the audition pool is not your core audience. It is people who have never heard of you.

Here is the example that makes this concrete. If you teach Instagram growth, a core-audience topic would be: “How to warm up DM leads using a three-step voice note framework.” Your committed followers get it immediately. A stranger in the audition pool has no idea what “warm up DM leads” means. They scroll.

The middle circle is your casual audience. They know your niche. They have the pain point. They just are not specifically following you yet. A casual-audience topic might be: “Why your Instagram DMs are not converting and what to fix.” More people can find an entry point there. It still requires some context though.

The outer circle is your new audience. These people have a universal frustration. They may not be deep in your niche yet. A new-audience topic is: “Why you are posting every day and still not growing.” Anyone who has ever tried social media understands that immediately. No prior knowledge required.

Read that again.

The new-audience circle is the only one where cold strangers find an entry point without already knowing you. It is the only tier that consistently passes the audition.

But Max, doesn’t that mean I have to water everything down?

No. It means you lead with the universal frustration and then teach the mechanism underneath it. The mechanism does not change. The door does.

If you are in the fitness niche, a core-audience topic is “how I periodize carb cycling during a 12% cut.” Super duper key for your diehard followers. But “why you are not losing weight even though you are eating at a deficit” is something any person who has ever tried to lose weight understands in two seconds.

Same expertise. Different entry point. Boom.

If you are a business owner, this applies to you especially. Content about your specific internal process and your proprietary frameworks is almost all core-audience content. It serves the people who already know you. It does not grow your account.

What Signals Does the Instagram Algorithm Actually Look At?

Instagram head Adam Mosseri confirmed three top signals in January 2025: watch time, likes per reach, and sends per reach. These apply across Reels, Feed, and Explore, weighted differently per surface. For reaching cold audiences through Reels, sends per reach is the most powerful of the three.

Watch time

This is the one everyone knows. And yes, it matters.

But most people think about it slightly wrong. Mosseri clarified in February 2025 that Instagram does not only look at the percentage of the video someone watched. It also counts total seconds including replays. A 15-second Reel that most people watch twice beats a 60-second Reel that most people finish once.

A replay is not just a nice metric. It is one of the highest signals you can generate.

When a video stalls between 100 and 900 views, this is what happened: the audition pool watched maybe 20 to 30% of it. That fell below the threshold for expansion. Distribution stopped.

Likes per reach

Likes are a connected-reach signal. They tell the algorithm your existing followers enjoyed the content. That is useful for keeping your current audience engaged. But likes from followers do not trigger cold discovery. They do not push you into Explore or into the feeds of people who have never heard of you.

Sends per reach

This is the signal almost nobody is talking about.

Sends per reach – the ratio of DM shares to total views – is weighted 3 to 5 times higher than likes when the algorithm calculates non-follower distribution. A rate of 1 to 2% is considered strong. Above 3% is viral tier.

Think about what a DM share actually is. Someone watched your video and immediately thought of a specific person who needed to see it. They opened a conversation and sent it. That is a deliberate, high-intent action. It only happens when a video hits a nerve broad enough to feel universal to people outside your existing circle.

Core-audience videos almost never get sent. New-audience videos get sent all the time. Pretty straightforward when you see it that way.

Why Does Who Shares Matter as Much as How Many Shares?

The algorithm does not just count how many times your video was DM’d. It tracks who those sharers are. Shares from cold viewers outside your follower base carry significantly more weight for new-audience distribution than shares from your own followers. A connected share and an unconnected share are not the same signal – and the algorithm treats them very differently.

Most creators know shares are good for reach. But almost no published article explains this layer underneath.

A share from someone who already follows you is weighted lower for new-audience distribution than a share from a cold viewer who found the video through the audition pool.

But wait. A share is a share. Why does the algorithm care who did it?

Because the algorithm is trying to figure out whether your video is interesting to people who have never seen you before. A share from your own follower tells it your existing audience likes you. Useful. But not a signal about new-audience relevance. A share from a cold viewer who found the video through the audition pool tells it something completely different: this content is relevant to people who do not already know this creator.

That is the signal that triggers wide distribution.

This is why a new-audience topic can outperform a more-engaging core-audience video in raw reach, even when the core-audience video has stronger follower metrics. The shares coming off the new-audience video are from cold viewers. The algorithm weights those more heavily for Explore and discovery placement.

What Is the Connected vs. Unconnected Reach Split?

Connected reach is distribution to your existing followers, driven primarily by likes per reach. Unconnected reach is discovery distribution to strangers, driven primarily by sends per reach. These are two separate systems and a video can perform strongly in one while getting zero traction in the other at the same time.

The connected-reach system keeps your current followers engaged. That matters. But it does not grow your account.

The unconnected-reach system is the growth engine.

A video can score strongly on connected reach – high likes from followers, excellent watch time from people who know your work – and score zero on unconnected reach in the same posting cycle. If you look at your Insights and see solid follower engagement but flat month-over-month reach, this is almost certainly what is happening.

Your content is excellent at retaining the audience you have. It is not passing the cold audition.

About 55% of Reel views come from non-followers for accounts consistently hitting the unconnected-share signal. If you are deep in core-audience content, you are likely well below that number. The fix is not a better hook. The fix is a different topic.

What Changed in the Instagram Algorithm in 2025 and 2026?

In the past year, Instagram added two-way creator-viewer conversation as a ranked signal, made Trial Reels schedulable, launched topic controls that let users choose what they see, and confirmed that hashtags do not improve reach. An ongoing originality penalty also removes heavy reposters from recommendations entirely.

Here is each update in plain terms.

Two-way conversation signal (March 2026)

In March 2026, Instagram added creator-viewer conversation as a ranked signal. If a viewer DMs you and you reply within one hour, that exchange compounds your reach ranking. Comment threads between you and a specific viewer are weighted similarly.

This is new. Most published articles about the algorithm have not caught up to it yet. Reply to DMs quickly – not as a relationship habit, but as a reach strategy.

Trial Reels (schedulable as of February 2026)

Trial Reels became schedulable in February 2026. This feature posts your Reel to non-followers only for the first 24 hours. If it performs well in that window, Instagram then distributes it to your followers too.

This is a built-in audition tool. Use it to test new topic angles before committing them to your main feed. You get real cold-pool data without risking your follower engagement numbers.

“Your Algorithm” controls (December 2025)

Instagram launched a feature called “Your Algorithm” in December 2025. It lets users toggle topic categories on and off in their discovery feed. They actively choose what they want to see.

The implication for creators: your content needs to match a recognizable topic category that your target viewers have opted into. Highly niche content that does not map to a standard category is harder to surface through discovery than it was before.

Hashtags (confirmed cap: five)

Mosseri has said publicly that hashtags do not improve reach. The confirmed cap is five per post, used for categorization only. Any article telling you to use 20 to 30 hashtags is out of date.

Originality penalty (ongoing)

Accounts that repost more than 10 times in a 30-day window are excluded from Instagram’s recommendation surfaces entirely. Watermarked TikTok clips trigger the same penalty. Original content gets roughly 40 to 60% more distribution than reposts. If you are repurposing content from other platforms, remove any watermarks and reformat for Instagram before posting.

How Do I Actually Fix My Reach?

Start with the CCN audit. Run your last ten videos through one question: could a complete stranger find an entry point in the first five seconds? Most creators find that seven or eight out of ten videos are core-audience content. That single audit tells you more about your stalled reach than any analytics dashboard ever will.

From there, five steps in order.

Step one: run the CCN audit.

Look at each of your last ten videos. Ask honestly: core, casual, or new? If you are freaking certain a stranger could walk in on that video and immediately understand why it is relevant to them, that is new-audience. If it requires context or prior knowledge, it is core.

Step two: write a new-audience version of every core-audience topic you want to make.

Take the specific mechanism and find the universal frustration underneath it.

“My three-step DM conversion framework” is core audience. “Why your DMs never turn into clients” is new audience. Same expertise. Different door.

Step three: start tracking sends per reach.

In your Insights, find how many times each video was sent via DM and divide that by total plays. If that number is consistently below 1%, your topics are landing with followers but not with cold viewers. You are winning connected reach only.

Step four: use Trial Reels to test new topic formats.

Get real cold-pool data before committing a new format to your main feed.

Step five: reply to every DM within the first hour.

This is now a ranked signal. Treat it like one.

But Max, this is a lot. Which step actually matters most?

Step one. Just do the audit. Be honest. Once you can clearly see that most of your content is aimed at people who already know you, everything else follows. That single shift in how you pick topics is where the answer lives for most stalled accounts.

What About Just Writing Better Hooks?

A hook fix only works when strangers are bouncing because the opening line is weak. If strangers are bouncing because the topic has no entry point for them, rewriting the hook changes nothing. Fix the topic first. The hook is a second-order problem.

A better hook helps when strangers are leaving in the first two seconds. That is a real problem and worth solving. But it is the wrong diagnosis when the actual issue is that strangers understand the topic immediately and know it is not for them.

Here is how to tell which problem you actually have. Look at your Reel Insights.

If watch time is high among followers but reach to non-followers is flat, the topic is the problem. Your followers are watching to the end. Strangers are bouncing as soon as they understand what the video is about.

If watch time is low across the board – followers and non-followers both bouncing early – that is a hook problem.

Fix the topic first. Then fix the hook. In that order.

Once you are consistently picking new-audience topics, hook craft becomes a real lever. But polishing a hook on a core-audience topic is like polishing an entrance to a room strangers were never going to walk into in the first place. I digress. You get the gist.

Are You Building the Right Audience?

Getting cold viewers to watch your Reels is one problem. Turning that attention into something that actually matters for your business is a different one entirely.

The algorithm gets strangers to the door. What you teach them and what you invite them toward determines whether that audience does anything for you beyond a follower count. If you want to understand how to move the people you are attracting from follower to client, how to turn Instagram followers into clients covers the framework for that next step.

---

Want the templates behind content that actually gets watched? Grab the templates here and start with the first one today.