Glossary
Engagement Weighting
Not all comments are equal. Likes and replies reveal which ones matter.
Definition
Engagement weighting adjusts the influence of each comment in an analysis based on the number of likes and replies it received. Comments with high engagement shaped the visible conversation; comments with zero engagement did not. Weighting ensures the analysis reflects what the audience actually amplified.
How it works
- 1
Each comment's weight is calculated: log(likes + 1) + (replies × 0.5).
- 2
Log scale prevents a single viral comment from dominating the results.
- 3
Replies are weighted at 0.5× because replying is higher-intent than liking.
- 4
Stance distributions and narrative scores are recalculated using weighted comment counts.
- 5
The result is a weighted stance split that better reflects actual audience influence.
Why it matters
A raw comment count treats a zero-like comment the same as one with 5,000 likes. In practice, the high-engagement comment was seen and amplified by thousands of people; the zero-engagement comment was not. Engagement weighting aligns the analysis with the conversation that actually happened publicly.
Related distinctions
Engagement weighting vs raw comment count
Raw comment counts give an unweighted picture of who commented. Engagement-weighted counts give a picture of whose comments shaped the conversation. For PR and public affairs analysis, the weighted view is typically more relevant.
Engagement weighting vs reach data
Reach data (impressions, views) measures content exposure. Engagement weighting measures comment-level influence within the audience that chose to interact. Both are useful but answer different questions.
Frequently asked questions
What is engagement weighting in comment analysis?
Engagement weighting assigns a numeric score to each comment based on its likes and replies, then uses those scores to adjust stance distributions and narrative rankings. High-engagement comments count more because they demonstrably shaped the visible conversation.
Why use a log scale for engagement weighting?
A log scale prevents a single viral comment (e.g. 50,000 likes) from overwhelming the entire analysis. It compresses the high end so that a comment with 100 likes still contributes meaningfully alongside one with 10,000.
Why are replies weighted differently from likes?
Liking a comment takes one tap. Replying requires writing a response — it is a higher-intent signal of engagement. Replies are weighted at 0.5× rather than 1× to reflect this difference without over-counting them.
Does engagement weighting change the stance split significantly?
It often does, especially for polarising posts. A post where the top-liked comments are strongly supportive can show a very different weighted stance split compared to its raw split. That difference is informative — it tells you which side amplified more effectively.
See engagement weighting in practice
Narativ applies stance analysis, narrative clustering, and engagement weighting to live comment sections — from £1 per post.