If your views fell while your content stayed the same, the answer is usually in how Instagram assembles a feed and not in anything you did wrong.

Meta publishes the mechanics. Its Transparency Center system card for the Instagram Feed AI system lays out a four stage pipeline that starts with every post from the accounts you follow and ends with a few hundred of them in a scored order.

Two numbers from that card frame the whole problem. The system shortlists approximately 700 posts, then scores about 500 of them. Most of what creators call the algorithm is really the fight to survive those two cuts.

How Instagram Actually Builds Your Feed

Instagram ranks your feed in four stages that Meta names directly: gather available posts, leverage signals, make predictions, and rank posts by score. This card covers the connected feed, meaning posts from accounts you already follow. It is not the recommendation system behind Explore or suggested posts, so do not read these numbers as the rules for reaching strangers.

Stage one gathers “all potential posts from accounts you follow” and strips out anything breaching the Community Guidelines. Stage two runs what Meta calls “a lightweight model” to “select approximately 700 of the most relevant posts”.

Stage four then “calculates a relevance score for about 500 posts and puts them in order by this score”. That scoring pool is the actual competition your post enters, and it is rebuilt for every viewer instead of being decided once globally.

What the Ranking Signals Reward

The card lists input signals feeding those predictions, and they are more specific than the usual advice about hooks and consistency. Several have measurable definitions attached.

• Whether you “open the post’s share sheet and select someone to send it to”, which is the private share and not the public comment
• “How many times you’ve tapped the reshare button in the last 28 days”, a signal about the viewer’s own habits, not the post
• Whether the post is “informative content (news, product reviews, tutorials, how-tos)”
• “Whether it’s a photo and how much text is overlaid on it”
• “How likely you are to skip this post compared to other posts shown alongside it”

That last one matters more than its wording suggests. Your post is measured against whatever else landed in the same viewer’s scoring pool that day, so the bar moves with the competition.

Instagram also runs several separate predictions about the very first post in a session, including whether you skip it, whether you stay on it past ten seconds, and whether it keeps you spending time in your feed for the rest of that session. The first slot is treated as its own problem.

Why Your Reach Dropped Even if Your Content Did Not

Reach can fall without any drop in quality, because the scoring pool is competitive and built per viewer. Here is what genuinely moves it.

• The accounts each person follows are posting more, so the shortlist of roughly 700 is drawn from a bigger candidate pool and more of your posts miss the cut
• Your followers’ own behavior is an input. Several signals describe what the viewer does, such as their reshare habit over the last 28 days, so a passive audience weakens your reach without you changing a thing
• Feed composition is deliberately mixed. Meta states the system “tries to ensure your feed has a balanced mix of content types”, which caps how much of any one format a single account can occupy
• Meta keeps changing the ranking models, and says so on the card, so a format that worked one month can settle differently the next

Your analytics will not label any of this. You see a quiet decline you cannot trace to a specific post, which is why so many creators reach for a shadowban theory instead.

Does Your Account Type Actually Matter?

Not for ranking, as far as Meta’s own documentation goes. The Feed system card publishes what it calls some of the significant signals its ranking models take in, and account type does not appear among them. Neither does follower count, nor a verification badge.

Hold that at the right strength. Meta calls its own list partial and says these models and their input signals change frequently, so the absence of account type is not proof that it counts for nothing. The card does show where the published pipeline puts its attention, which is on the viewer’s predicted response to a post.

Account types do differ, though not in the way the usual advice claims. Switching to a Creator or Business account changes which tools and analytics you unlock, while distribution stays with the ranking models either way.

• Creator and Business accounts unlock deeper insights, scheduling, branded content tags and ad tools
• Business accounts add contact buttons, shopping features and full ad manager access, which is why brands use them
• Personal accounts keep it simple, and are more limited on analytics while facing the same follow limits everyone does

If your reach fell right after you switched account types, the switch is the most visible change but rarely the cause. Look at what happened to your posting rhythm and format mix in the same period, and at whether anything triggered an account restriction.

Carousels, Reels, or Photos: What Actually Works

No format is switched off. The card gives carousels their own dedicated prediction, “how likely you are to swipe through the whole carousel”, with signals for how many cards you view and how many swipes you make, which is not how a sidelined format gets treated.

Format still shapes the prediction. “Whether the post is a photo, video, or carousel” is an explicit signal, and so is how much text sits on top of a photo.

• Reels are where Instagram keeps investing, and are widely treated as the format that travels furthest beyond your followers. The Feed card does not cover that, since it describes only the feed built from accounts you already follow
• Carousels are scored on completion. A carousel people swipe to the end of is doing exactly what its prediction measures, which rewards a real sequence over a dump of images
• Single photos are not penalized, but they compete in a mixed feed with less to measure, and heavy text overlay is itself a tracked signal

Instagram is actively varying formats in each viewer’s feed, so an account posting only one format is competing for a narrower slice of it.

How to Adapt Without Burning Out

Work with the parts of the system that are documented and stable, instead of chasing month to month theories. The signals above point at a short list of things worth your effort.

• Make posts worth sending. Sending a post in a direct message gets its own dedicated prediction on the card, while liking appears only as one example inside a broader positive-action signal
• Batch film Reels in advance, and study what makes a Reel travel before chasing volume
• Build carousels people finish, since completion is what the carousel prediction measures
• Reply to comments and DMs. Your audience’s activity level feeds the signals, so an engaged following protects your reach
• Vary your formats instead of committing to one, given the feed is assembled as a mix
• Use tools like Flick to plan your calendar and track what is actually landing
• If you are selling something behind the content, keep the funnel separate with Systeme IO

Reach that falls for structural reasons responds to giving the ranking models something they actually measure, which is a post people finish and send on. Volume alone will not move it. If reach stays flat across every format for weeks, read up on what organic reach on Instagram looks like now before assuming your account is being suppressed.

Frequently Asked Questions

Does your Instagram account type affect your reach?

Not according to Meta’s published documentation. The Instagram Feed system card publishes what it calls some of the significant input signals its ranking models use, and account type does not appear among them. Meta describes that list as partial, so treat it as what Meta has disclosed about Feed ranking rather than a complete account of every signal.

How many posts does Instagram actually rank for your feed?

Meta’s card describes a lightweight model selecting “approximately 700 of the most relevant posts” from the accounts you follow, then a scoring stage that “calculates a relevance score for about 500 posts”. Those figures apply to the connected feed, not to Explore or recommendations.

Are photos and carousels deprioritized compared to Reels?

The card does not describe any format being switched off, and carousels get their own dedicated prediction covering how far you swipe through them. Meta also states the system “tries to ensure your feed has a balanced mix of content types”. Reach beyond your own followers runs on a different system that this card does not cover.

Does being verified help your posts rank?

Verification is not among the signals listed on the Feed system card. Verified accounts are often large accounts with active audiences, and audience behavior is a real signal, so the visibility gap people notice has an explanation that does not require the badge itself to count.

Why did my reach drop when my content did not change?

Ranking is per viewer and competitive. Your post is scored against whatever else reached the same person’s shortlist, and several signals describe the viewer’s own habits rather than your post, so reach can move when nothing about your content did.

What does a balanced mix of content types mean?

It is Meta’s own phrase for the final ranking stage, where the system “tries to ensure your feed has a balanced mix of content types” instead of filling a feed with one format. In practice it limits how much of any single viewer’s feed one format or one account can take.