The Art and Science of Visibility: Navigating Instagram’s Algorithm
How Instagram Ranking Algorithm(s) Work.
1. FEED — Feed Ranking Process
Algorithm: Personalized Recommendation Algorithm
Definition: A personalized recommendation algorithm analyzes user interactions, content information, and user history to make educated guesses about the user’s preferences and rank the content accordingly, aiming to maximize user engagement and satisfaction.
To maximize the probability of success on the platform, the algorithm could consider the following data points and assign appropriate weights to each:
User Activity:
Weight: w1
Definition: User activity refers to the actions taken by the user, such as likes, shares, saves, and comments on posts. It provides insights into the user’s interests and preferences.
Formula: UserActivityScore = w1 * (Number of Likes + Number of Shares + Number of Saves + Number of Comments)
Post Information:
Weight: w2
Definition: Post information includes signals related to the popularity of a post (e.g., likes, comments, shares) and content attributes (e.g., post time, location, attached tags). It helps evaluate the overall appeal and relevance of the post.
Formula: PostInformationScore = w2 * (PopularityScore + ContentAttributesScore)
Person Information:
Weight: w3
Definition: Person information encompasses signals about the person who posted the content, including their past interactions, engagement, and overall interest to the user. It assists in assessing the user’s potential interest in the person’s posts.
Formula: PersonInformationScore = w3 * (InteractionsScore + EngagementScore)
History of Interactions:
Weight: w4
Definition: History of interactions captures the user’s previous interactions with a particular person’s posts, providing insights into the user’s general interest in seeing content from that person.
Formula: InteractionHistoryScore = w4 * (CommentHistoryScore + LikeHistoryScore)
Overall Ranking Score:
Ranking Score = UserActivityScore + PostInformationScore + PersonInformationScore + InteractionHistoryScore
2. STORIES — Stories Ranking Process
Algorithm: Personalized Recommendation Algorithm
Definition: A personalized recommendation algorithm utilizes user-specific data, such as viewing history, engagement, and relationship closeness, to predict and rank Stories that the user is more likely to find relevant and valuable, enhancing their overall experience on the platform.
To maximize the probability of success on the platform, the algorithm could consider the following data points and assign appropriate weights to each:
Viewing History:
Weight: w1
Definition: Viewing history refers to how often the user views stories from a particular account. It helps identify the accounts that the user doesn’t want to miss and prioritizes their stories.
Formula: ViewingHistoryScore = w1 * FrequencyOfViewing
Engagement History:
Weight: w2
Definition: Engagement history takes into account how often the user engages with stories from a specific account, such as sending likes or direct messages. It indicates the user’s interest and involvement with the account’s stories.
Formula: EngagementHistoryScore = w2 * FrequencyOfEngagement
Closeness:
Weight: w3
Definition: Closeness considers the overall relationship between the user and the story author, assessing the likelihood of being connected as friends or family. It helps prioritize stories from accounts that have a closer relationship with the user.
Formula: ClosenessScore = w3 * RelationshipCloseness
Overall Ranking Score:
Ranking Score = ViewingHistoryScore + EngagementHistoryScore + ClosenessScore
3. EXPLORE — Explore Ranking Process
Algorithm: Personalized Recommendation Algorithm
Definition: A personalized recommendation algorithm analyzes user-specific data, including past activity, post information, interaction history, and person information, to predict and rank content that the user is likely to find interesting and valuable, thereby enhancing their exploration experience on the platform.
To maximize the probability of success on the platform, the algorithm could consider the following data points and assign appropriate weights to each:
Information about the Post:
Weight: w1
Definition: Information about the post includes signals of its popularity, such as the number and speed of likes, comments, shares, and saves. These signals carry more weight in Explore compared to Feed or Stories.
Formula: PostInformationScore = w1 * (PopularityScore + EngagementScore + SaveScore)
User Activity in Explore:
Weight: w2
Definition: User activity in Explore refers to the user’s interactions with posts, including likes, saves, shares, and comments, as well as their overall engagement with Explore content. The algorithm aims to show more similar content based on the user’s interaction patterns.
Formula: ExploreActivityScore = w2 * (FrequencyOfLikes + FrequencyOfSaves + FrequencyOfShares + FrequencyOfComments)
History of Interactions with the Person Who Posted:
Weight: w3
Definition: The history of interactions with the person who posted the content provides insights into the user’s interest and engagement with that individual’s posts. It helps determine the likelihood of the user being interested in the content shared by that person.
Formula: InteractionHistoryScore = w3 * InteractionsScore
Person Information:
Weight: w4
Definition: Person information includes signals about the person who posted the content, such as the frequency of interactions with that person over the past few weeks. It helps identify compelling content from a diverse range of people.
Formula: PersonInformationScore = w4 * InteractionFrequencyScore
Overall Ranking Score:
Ranking Score = PostInformationScore + ExploreActivityScore + InteractionHistoryScore + PersonInformationScore
4. REELS — Reels Ranking Process
Algorithm: Personalized Recommendation Algorithm
Definition: A personalized recommendation algorithm analyzes user-specific data, including activity, reel information, and person information, to predict and rank Reels that the user is likely to find entertaining and engaging, enhancing their experience on the platform.
To maximize the probability of success on the platform, the algorithm could consider the following data points and assign appropriate weights to each:
User Activity:
Weight: w1
Definition: User activity refers to the user’s interactions with Reels, such as likes, saves, reshares, comments, and recent engagements. It helps understand the user’s preferences and relevance to specific content.
Formula: UserActivityScore = w1 * (LikesScore + SavesScore + ResharesScore + CommentsScore)
History of Interactions with the Person Who Posted:
Weight: w2
Definition: The history of interactions with the person who posted the Reel provides insights into the user’s interest and engagement with that individual’s content. It helps determine the likelihood of the user being interested in the Reel shared by that person.
Formula: InteractionHistoryScore = w2 * InteractionsScore
Information about the Reel:
Weight: w3
Definition: Information about the Reel includes signals about the content within the video, such as the audio track, visuals, and popularity indicators. These signals assist in assessing the entertainment value and engagement potential of the Reel.
Formula: ReelInformationScore = w3 * (AudioTrackScore + VisualsScore + PopularityScore)
Person Information:
Weight: w4
Definition: Person information encompasses signals about the person who posted the Reel, such as their popularity indicators (e.g., number of followers, engagement levels). It helps identify compelling content from a diverse range of creators and gives everyone a chance to find their audience.
Formula: PersonInformationScore = w4 * (FollowersScore + EngagementScore)
Overall Ranking Score:
Ranking Score = UserActivityScore + InteractionHistoryScore + ReelInformationScore + PersonInformationScore
5. Search — Search Ranking Process
Algorithm: Search Ranking Algorithm
Definition: The search ranking algorithm determines the order of search results based on their relevance to the user’s query. It aims to provide the most accurate and useful results by considering various data points.
Data Points and Weighting Strategy:
Query Relevance:
Weight: w1
Definition: Query relevance focuses on how closely the content matches the user’s search query. It considers factors such as keywords, tags, captions, and relevance algorithms to determine the relevance of a post to the search query.
Formula: QueryRelevanceScore = w1 * (KeywordMatchScore + TagMatchScore + CaptionMatchScore)
User Engagement:
Weight: w2
Definition: User engagement considers the popularity and engagement metrics of the content. It takes into account factors such as the number of likes, comments, shares, and views the content has received, indicating its relevance and value to users.
Formula: UserEngagementScore = w2 * (LikesScore + CommentsScore + SharesScore + ViewsScore)
Content Popularity:
Weight: w3
Definition: Content popularity focuses on the overall popularity or quality of the content. It takes into account factors such as the reputation or authority of the account or the content’s historical performance, aiming to showcase popular and high-quality content.
Formula: ContentPopularityScore = w3 * (AccountPopularityScore + ContentQualityScore)
User Preferences:
Weight: w4
Definition: User preferences refer to the user’s past interactions, interests, and behavior on the platform. It considers factors such as the types of posts they engage with, accounts they follow, or hashtags they interact with, helping to personalize the search results.
Formula: UserPreferencesScore = w4 * (PostInteractionsScore + AccountInteractionsScore + HashtagInteractionsScore)
Overall Ranking Score:
Ranking Score = QueryRelevanceScore + UserEngagementScore + ContentPopularityScore + UserPreferencesScore
Voila!
Well…it’s not much of a philosophy.
I know!
But well…#$@% &*^…Don’t come following me!
~The French Waiter