Course contentsShow
← System Design
Lesson 1666 of 1,91940. Real-World System: Social Media FeedFree lesson

Ranking Signals and Features

Key signals used in ranking algorithms including recency, engagement metrics, author relationship, and content type.

Ranking Signals and Features

What you'll learn: The key signals and features that ranking algorithms use to decide which posts appear first in a user's social media feed.

The Core Idea

When a social media feed has hundreds or thousands of potential posts to display, the system needs criteria to decide what users see first. Ranking signals are measurable properties or metadata about posts and users that indicate how relevant or interesting a post might be. These signals feed into machine learning models or scoring algorithms that produce a personalized feed order.

Think of it like a restaurant recommendation system: it doesn't just show you random restaurants—it considers distance (recency), how often friends ate there (engagement), who recommended it (relationship), and cuisine type (content type).

Primary Ranking Signals

Recency: How old is the post? Fresh content typically ranks higher because it's more timely and relevant. A post from 2 hours ago often matters more than one from 2 days ago.

Engagement Metrics: Likes, comments, shares, saves, and click-through rates. High engagement suggests compelling content that resonates with users. A post with 500 likes in 10 minutes signals viral potential.

Author Relationship: Your connection strength with the post creator. Posts from close friends, family, or accounts you frequently interact with rank higher. A best friend's post beats a distant acquaintance's.

Content Type: Photos, videos, links, or text posts. User preferences vary—some engage more with videos, others with articles. The system learns what you prefer.

Additional signals include user dwell time (how long you view), negative feedback (hides/reports), hashtag relevance, and historical interaction patterns.

Building the Feature Vector

These signals become features in a machine learning model—numerical inputs that predict engagement probability. The model might combine 100+ features to generate a single relevance score per post.

Key Takeaway: Ranking signals like recency, engagement, relationship strength, and content type transform raw posts into a personalized, scored feed that maximizes user engagement and satisfaction.