Content ranking

Instagram’s ranking systems work with different pools of content, from followed accounts’ updates to public posts selected for discovery.314 Instagram describes Feed as a mixture of posts from followed accounts, advertisements and recommended posts, whereas the Stories system gathers non-advertising stories from accounts the viewer follows.315 For Feed recommendations, Meta’s Transparency Center says the system gathers a portion of public photos and videos that meet its quality and integrity rules, then considers interests and engagement with similar content.14 For Stories, it considers who shared a story and the viewer’s previous interactions with that person, predicts personal value and applies additional checks before ordering the stories.3

For Explore ranking, Meta’s Transparency Center describes finding up to 1,500 possible photos and videos using sources including authors a viewer has engaged with and media similar to media they have engaged with, then narrowing those choices through ranking stages.16 Meta’s engineering team’s August 2023 account of Instagram Explore described final adjustments for predicted engagement, content integrity and diversity.17 Comment ranking has a different task: deciding the order of replies beneath a post rather than choosing the post itself.2 For comments, Meta’s Transparency Center says Instagram can place a viewer’s own comments, friends’ comments or pinned comments first, score other comments for factors including relevance to the post and popularity, and filter Community Guidelines violations.2

For creators, Instagram’s ranking explanations identify several kinds of evidence rather than a single engagement total.181920 Recommendations can bring creators’ content to people who do not yet follow them, helping them reach new audiences.421 Instagram’s Help Center says suggested posts draw on a viewer’s follows, likes, saves, reposts and comments, connections with similar accounts, post popularity, posting time or location, and recent interactions with the account.18 Instagram also says a prediction that someone will report a post can move it lower in that person’s Feed.15 Mosseri explained in January 2025 that likes provide more information for training ranking systems because they are more common than saves, while saves still help indicate interest.19 His August 2024 explanation gave a different reason for valuing sends: sharing something with a friend creates a social connection, whereas saving is usually an individual activity.20

Mosseri’s explanations also distinguish audience response from deliberate penalties and acknowledge limits in how ranking works.9222324 Mosseri said in May 2026 that identical Reels can receive very different reach because timing, competition and context change how people respond; in February 2026, he also attributed weaker ranking after a long posting break to having less information about the interested audience, not an intentional penalty.922 His February 2025 explanation distinguished photos and carousels performing less well from deliberate downranking, saying Reels tend to generate more engagement.24