Instagram is giving users something they have long lacked in the algorithmic feed — a direct way to tell the platform what they actually want to see. The company has expanded its Your Algorithm controls to the main feed, extending topic-level management tools that were previously available only for Reels and Explore. The move comes as Instagram continues its fundamental shift from a social graph built on follower relationships to an interest graph driven by engagement signals and machine learning.
The Your Algorithm feature, first introduced for Reels in December, lets Instagram users view, add, and remove the topics the platform associates with their interests. Until now, those controls were limited to the Reels surface. With this update, the same topic management applies to the main feed, where recommended posts from accounts users do not follow have become a dominant part of the experience. The change effectively gives users a dashboard of their own algorithmic profile — a list of content categories the system believes they care about — and the ability to edit that profile directly.
When users open the Your Algorithm controls, they see a list of topics Instagram has generated from their activity. These might include categories such as photography, cooking, running, interior design, or music production — any subject cluster the platform has inferred from taps, watches, shares, and time spent. Users can remove topics they find irrelevant or no longer want influencing their recommendations. They can also add topics they want to see more of, signalling intent rather than merely reacting to what the algorithm serves. Changes to the list help the system adjust future recommendations across Feed, Reels, and Explore.
The update addresses a structural tension at the heart of modern social platforms. For years, Instagram and its competitors have relied on implicit signals — the content users engage with — to train recommendation models. Those signals are powerful, but they are also indirect. A user who watches a cooking video may be interested in recipes, or they may have simply paused on a satisfying clip of knife skills. The algorithm has no way to distinguish passive consumption from active interest without explicit input. The Your Algorithm controls provide that explicit input, turning recommendation tuning from something the platform does to users into something users do with the platform.
The Mechanics of Topic-Level Control
The underlying technology that makes this possible has shifted considerably in the past two years. Instagram head Adam Mosseri explained that large language models now allow the platform to describe content clusters in plain language, rather than relying on opaque machine labels or broad category tags that users would not recognize. Instead of seeing something like “Category 47: high-engagement food content,” a user sees “cooking” or “baking” or “sourdough.” This semantic understanding, powered by LLMs, bridges the gap between the platform’s internal content taxonomy and the intuitive categories users naturally think in.
Mosseri framed the update as a response to a persistent user complaint. “The system learns from what you tap, watch, and share, but you don’t really get to tell it what you want,” he wrote. That distinction — between what the system infers and what the user declares — is central to the design of the new controls. The topic list is not a replacement for the algorithmic signals Instagram already uses. It is an overlay, a set of explicit parameters that sit alongside the implicit signals the system continues to collect. When a user removes a topic, they are not telling Instagram to ignore that content entirely. They are telling the platform to deprioritise it in the mix of signals that determine recommendations.
This hybrid approach has practical consequences for how recommendation quality evolves. A user who follows several fitness accounts but is currently training for a marathon can add “marathon training” as a topic, shifting the algorithm’s weighting toward long-distance running content without losing the broader fitness recommendations they still want. The system does not reset. It recalibrates.
The Broader Shift from Social Media to Interest Media
Gary Vaynerchuk has described this industry-wide transition as a movement from social media to interest media. The phrase captures a fundamental reorientation of platform architecture. Traditional social media was built around the follower graph: users saw content from accounts they chose to follow, and the network’s value came from those explicit social connections. Interest media, by contrast, prioritises content based on engagement signals rather than follow relationships. A video about woodworking can appear on the feed of a user who has never followed a woodworking account, as long as the platform’s model predicts high interest based on that user’s broader behaviour.
Instagram’s update makes that model more transparent. Instead of wondering why a particular post appeared on the feed, users can inspect the topics the platform believes they care about and correct any misalignments. A user who sees “politics” on their topic list but prefers to avoid political content can remove it directly. A user who wants more landscape photography but rarely encounters it can add the topic and signal that preference. The control panel acts as a window into the algorithm’s reasoning, giving users a degree of insight that was previously unavailable.
This transparency serves Instagram’s interests as well as users’. When recommendation quality improves because users have explicitly shaped their topic profiles, engagement metrics tend to follow. Users who feel they can control their feed are less likely to experience the fatigue that comes from irrelevant recommendations. They scroll longer. They interact more. The platform’s core engagement loop strengthens.
What This Means for Content Creators and Marketers
The expansion of Your Algorithm controls carries significant implications for anyone producing content on Instagram. Discovery on the platform increasingly prioritises matching user interests over follower relationships. That means content must signal clear topics and audience intent, because those signals help determine where recommendations appear. A post that is visually compelling but topically ambiguous may struggle to surface in the right recommendation clusters. A post that clearly belongs to a well-defined topic category — and signals that category through captions, hashtags, audio choices, and visual cues — is more likely to be placed in front of users who have expressed interest in that topic.
This shift rewards clarity and specificity. A food photographer who posts a recipe benefits not just from the quality of the image but from the strength of the topic signal. If the platform correctly associates the post with “baking” and the user has added “baking” to their interest list, the recommendation probability increases. Conversely, content that falls into a topic category the user has explicitly removed faces a steep disadvantage, regardless of its quality or the creator’s follower count.
For marketers, the strategic takeaway is straightforward: topic taxonomy matters. Understanding the categories Instagram’s language models use to classify content — and deliberately aligning content with those categories — becomes part of the distribution strategy. That may mean using more descriptive captions, selecting audio tracks that reinforce the topic signal, and structuring visual content to be clearly categorisable by machine vision models.
How the Feature Works in Practice
Accessing the Your Algorithm controls requires navigating to the settings within the Instagram app. Users can find the feature under the account settings menu, where a dedicated section for content preferences lists the generated topics. The interface presents topics as individual tiles or list items, each with an option to remove. A search or browse function allows users to add new topics that may not have appeared in the auto-generated list.
The auto-generated list evolves over time. As users engage with new content, the platform updates its topic associations. A user who starts watching gardening content after years of ignoring it will eventually see “gardening” appear on their list. Removing a topic does not guarantee that related content will never appear — the algorithm may still surface it based on other signals — but it does reduce the weight of that topic in the recommendation mix.
The feature is rolling out globally to all Instagram users. It does not require a specific app version update beyond the standard release cycle, though users may need to restart the app or refresh their settings to see the new controls appear.
What Are the Your Algorithm Controls on Instagram?
The Your Algorithm controls are a set of topic management tools that let Instagram users view, remove, and add the interest categories the platform uses to personalise recommendations. Instagram generates an initial list of topics based on each user’s activity — the accounts they follow, the posts they like, the Reels they watch, and the content they share. Users can delete topics they do not want influencing their feed and add topics they want to see more of. The controls currently apply to the main feed, Reels, and Explore, making them the most comprehensive recommendation management system Instagram has released to date.
Why Topic Controls Matter More Than Ever
The timing of this expansion is not accidental. Instagram has been steadily increasing the proportion of recommended content in the main feed for over a year. Posts from accounts users do not follow now make up a substantial share of what appears on the home screen, a shift that mirrors the approach TikTok popularised and that every major social platform has since adopted. As the ratio of recommended content to followed content has risen, so has user frustration with irrelevant or unwelcome recommendations.
The Your Algorithm controls give users a mechanism to address that frustration directly, without resorting to the nuclear option of muting or blocking entire accounts. A user who sees too many travel posts but follows several travel accounts can remove “travel” from their topic list and reduce the algorithmic weight of travel content, without unfollowing the people they actually want to stay connected to. That granularity is the key innovation. Previous controls — such as the “Not Interested” button on individual posts — were reactive and transient. Topic removal is proactive and persistent.
Mosseri has indicated that topics are only the beginning. He said Instagram is developing controls for people, moods, content types, and other signals. The implication is that Instagram envisions a future where users can fine-tune their feed along multiple dimensions simultaneously: not just what the content is about, but how it makes them feel, who created it, and what format it takes. A user might one day be able to say they want more calming landscape videos from small creators, fewer loud commentary clips from large accounts, and no political content at all — all through a unified control interface.
Challenges and Limitations
The Your Algorithm controls are not a cure-all. They operate within the boundaries of Instagram’s content classification system, which is imperfect. Large language models can describe content clusters with impressive accuracy, but they still make errors. A video about baking a cake might be classified as “cooking content” when the user actually wants to classify it as “dessert content” specifically. The granularity of the topic list depends on the sophistication of the underlying model, and users may find that certain niche topics are not available to add.
There is also the question of how topic removals interact with other recommendation signals. If a user removes “running” from their topic list but follows several running accounts and frequently engages with running posts, the algorithm faces a conflicting set of signals. Instagram has not disclosed how it resolves such conflicts, and users may find that topic removal has less impact than expected when other signals are strong.
Additionally, the controls require active maintenance. A user who adds topics when they first set up the feature but never returns may find that their topic list has drifted as their behaviour changes. The feature is most valuable when treated as an ongoing tuning tool rather than a set-and-forget configuration.
The Competitive Landscape
Instagram’s move places it ahead of some competitors on recommendation transparency but behind others in certain dimensions. TikTok has long offered a “Not Interested” long-press gesture and allows users to indicate they want to see less of a specific creator or sound, but it does not provide a central topic dashboard comparable to Your Algorithm. YouTube has its “Not interested” and “Don’t recommend channel” options, along with a more granular control panel for ad topics, but its recommendation controls for organic content are less consolidated than Instagram’s new system.
Meta appears to be betting that giving users more control over their algorithmic profile will improve satisfaction without reducing engagement — a balance that has traditionally been difficult to strike. If the bet pays off, Instagram may set a new standard for recommendation transparency across the industry, prompting competitors to develop similar topic-level controls of their own.
The next frontier, as Mosseri hinted, is mood-based and context-based controls. A user might want funny content in the evening and educational content in the morning, or relaxing content on weekends and high-energy content during the workday. Those temporal and emotional dimensions are considerably harder to model than static topic categories, but the progress Instagram has made with LLM-driven topic classification suggests the company sees a path forward.
The expansion of Your Algorithm controls to the main feed marks a meaningful step in the evolution of how users relate to algorithmic recommendations. For years, the black box of the feed has produced content that sometimes delights and sometimes baffles, with no clear mechanism for users to explain what went wrong. The new controls do not fully open that black box, but they add a window — and, more importantly, a steering wheel. Users can now tell Instagram what they want, rather than relying on the platform to guess. In an ecosystem increasingly defined by the quality of algorithmic recommendations, that shift is significant. The recommendations are not going away. But for the first time, users have a clearer say in where they lead.
