Nobody outside Meta knows Instagram’s ranking function. Not agencies, not consultants, not us. What exists is a set of statements Instagram has published, a larger body of things creators report, and a lot of confident guessing. This page keeps those three apart.
The short answer: Instagram does not have one algorithm. It runs a different ranking system for Feed, Stories, Explore and Reels, and each one predicts a different set of actions.
What Instagram says it optimises for
In its ranking explainer, Instagram’s head, Adam Mosseri, writes that “Instagram doesn’t have a singular algorithm that oversees what people do and don’t see on the app.” Each surface is tuned to how people use it: Stories for people you already know, Explore for discovery of accounts you do not follow, Reels for entertainment.
The underlying method across Meta’s surfaces is the same one described in Meta’s Transparency Center: gather eligible inventory, read signals, run predictions of what you are likely to do, then sort by a relevance score. Meta says it combines over 100 prediction models to produce that score.
Note: Figures here were verified against Instagram’s ranking explainer on about.instagram.com and Meta’s Transparency Center as of September 2026. Platforms change these without notice.
What signals does Instagram rank on?
Ordered as Instagram itself presents them, per surface.
| Surface | Signals Instagram names, in its own order | Optimising for |
|---|---|---|
| Feed | Your activity (likes, saves, shares, comments); post information (popularity, recency, location); poster information; your interaction history with that poster | What you engage with |
| Stories | Viewing history; engagement history; closeness of the relationship | People you know |
| Explore | Post information, with popularity weighted heavily; your Explore activity; your history with the poster; poster information | Discovery |
| Reels | Your activity (likes, saves, reshares); your history with the poster; reel information (audio, visuals, popularity); poster information | Entertainment |
Two things are worth noticing. Popularity is weighted more heavily on Explore than on Feed, which is why a post can flop with followers and still travel. And “closeness” appears only for Stories, which is why Stories reach tracks your real relationships rather than your posting cadence.
What changed most recently
Stated by Instagram: the surface-specific model above is the published position, dated May 2023 and still the live explainer.
Widely reported, not confirmed in writing: through 2025 and 2026, Mosseri has repeatedly told creators in video posts that sends (shares to another person via DM) are the signal he watches most, alongside watch time and likes, both measured per reach rather than in absolute numbers. This is consistent and comes from him directly, but it is video commentary rather than published documentation, so treat the exact weighting as unverified.
Practitioner heuristic, not from Instagram: specific numbers you see quoted for how much more distribution original content gets, or a repost count that triggers exclusion from recommendations, have no first-party source. Instagram has said it prefers original content. It has not published a multiplier.
Myths this kills
Shadowbans. Instagram has never confirmed a mechanism called a shadowban. What it does confirm is reduced distribution: Meta publishes the categories it demotes, including clickbait, engagement bait, fact-checked misinformation, and content likely to violate its Community Standards. That is a real, documented thing with a real cause. A sudden reach drop is far more often a format change, a seasonal dip, or a post that simply did not hold attention. The Account Status screen in the app is the only authoritative check. Our glossary explains the distinction in more detail on what a shadowban actually is.
Engagement pods. Unfounded as a growth strategy, and actively risky. Instagram’s stated signals are per-viewer predictions. A ring of accounts that likes everything teaches the model that those accounts like everything, which is close to no information. Meta separately demotes engagement bait.
“The algorithm hates links.” There is no published statement that a link in a caption or bio suppresses a post. Meta demotes clickbait links, which is a judgment about the destination and the framing, not about the presence of a URL.
Posting time is decisive. It is not. Timing shifts who is awake when your post first circulates, which matters at the margin. It cannot rescue a post nobody wanted. Our best time to post on Instagram guide says the same thing, and the Instagram posting time tool converts the general windows into your own timezone. Frequency matters more than the hour, which is the subject of how often to post on Instagram.
What actually moves distribution, in priority order
- Whether people finish and re-watch. Watch time is the one signal every credible account of Reels ranking agrees on.
- Whether people send it to someone. A send is a strong, costly action, and it puts your post in front of a person who did not choose to follow you.
- Format fit for the surface. A carousel that would rank fine in Feed is not a Reels candidate. Pick the surface first, then make the thing.
- The first two seconds. Everything downstream is conditioned on people not scrolling past.
- Saves and comments. Real signals, but weaker per unit than sends in everything Mosseri has said publicly.
- Consistency. Not a ranking signal in any published list. It matters because it gives the system more chances to find your audience.
What to do
Choose the surface deliberately. Write a hook that survives a scroll. Make something a person would forward to one specific friend, because that is the action the system appears to value most. Then repeat it enough times that a single bad post is noise rather than a verdict.
Scheduling Instagram alongside everything else
None of this is a scheduling problem, but consistency is the one lever on the list that a queue genuinely helps with. BulkPublish schedules Instagram feed posts, carousels, Reels and Stories from one calendar with 15 platforms, so a week of posting is planned once rather than negotiated daily. The scheduling feature page covers how the queue works.
The parts you control
You do not control the ranking function, and neither does anyone selling you a course about it. You control three things: the format you publish in, the hook that decides whether anyone stays, and whether you keep going long enough for the sample size to mean anything. Everything else on the internet about the Instagram algorithm is a description of those three, dressed up.