What Actually Drives Social Media Engagement: Platform Data Reveals Surprising Truths

By shrijeetverma13 · July 21, 2026

This analysis of 2,200 social media posts across platforms reveals that verification status, follower count, and trending status have almost no impact on…

Twitter has the highest average engagement score at 0.519, while YouTube leads in total interactions with over 10.5 million likes, comments, and shares combined.

YouTube has the highest average toxicity score (0.505), while Reddit leads in spam rate at 51.32%. Looking at topics, Health content shows the highest average toxicity (0.53), and Finance-related posts have the highest spam rate at 55.11%.

Neither verification status nor follower count meaningfully predicts engagement score. Verified users average 0.515 engagement versus 0.502 for non-verified users—a negligible difference. Follower count shows almost zero correlation (-0.009) with engagement score, meaning having more followers doesn't translate to higher engagement.

Sentiment scores vary noticeably across topic categories. Health posts show the most positive perception (average sentiment of 0.067), while Climate posts are perceived most negatively (average sentiment of -0.033), creating a perception gap of 0.1 points. Finance stands out as the most polarized topic, with the widest spread of opinions (standard deviation of 0.576), suggesting audiences are deeply divided on this subject compared to others.

Across 24 months from January 2024 to December 2025, engagement metrics stayed relatively stable with a slight upward trend overall. Average likes started at 9,266 per post and ended at 8,814, while the average engagement score rose from 0.50 to 0.54. Likes peaked at 11,143 in September 2025, and the engagement score hit its highest point (0.57) in June 2025. Comments declined somewhat over the period (from 2,521 to 1,311), while shares held steady or slightly increased (1,353 to 1,439). A chart tracking likes, comments, shares, and engagement score over time, along with supporting data tables, was generated to illustrate these trends.

Based on the available data, trending topic posts do not show a significant boost in engagement. Trending posts averaged 9,984 likes compared to 10,032 likes for non-trending posts, a difference of about -0.5%, meaning there was essentially no meaningful increase in likes for trending content.

The correlation matrix compared seven metrics: follower count, likes, comments, shares, engagement score, sentiment score, and toxicity score. Overall, the relationships between these variables were quite weak. The strongest positive link was between like count and comment count (correlation of 0.040), while the strongest negative link was between engagement score and sentiment score (correlation of -0.033). A heatmap visualization was created showing the full correlation matrix across all seven metrics, making it easy to see how each variable relates to the others at a glance.

The stacked bar chart shows how emotions are distributed within each topic category as percentage shares. Fear is the dominant emotion for Climate, Health, and Sports topics, while Happy dominates Entertainment, Politics, and Technology. Finance stands out as the only category led by Anger. The strongest single emotional association across all topics is Happy in Politics, making up 26.16% of that category's posts.

Using the IQR method, no outliers were detected in post length, like count, or comment count — all values fell within the normal expected ranges (0% outliers in each column). Since there were no outlier posts to compare, the spam and trending rates shown reflect the overall dataset: about 50.86% of posts were flagged as spam and 47.55% were trending among the 2,200 posts analyzed.

The analysis ranked the top 10 languages by post volume and compared their average engagement and sentiment scores. A grouped bar chart and a bubble scatter plot were generated, letting you see side-by-side how each language performs on both metrics, with bubble size representing post volume in the scatter view. Detailed ranking tables were also produced showing each language's post volume, average engagement score, average sentiment score, and their respective ranks among the top 10.

The sentiment analysis of your post text shows an overall positive tone. Out of all posts analyzed, 59.4% were positive, 21.1% were negative, and 19.5% were neutral, with an average sentiment polarity score of 0.140 (leaning slightly positive).

A word cloud has been generated from the 'post text' column, visually displaying the most frequently used words in your text data. Words that appear more often are shown larger, making it easy to spot dominant themes and common terms at a glance.