Render-blocking JavaScript and CSS
Posted: Tue Feb 11, 2025 5:31 am
Methodology of the study
From our Google keyword database, we extracted 15,000 keywords that triggered a video to appear in SERP features. We then looked at the top ten YouTube search results for each of these keywords to understand how different video characteristics affect their positions in YouTube search. Based on this data, we gathered a wide range of metrics to assess which ones affect video performance.
In particular, we analyzed the performance of "how-to" keywords compared to general keywords, the positioning of videos in search engines based on their engagement metrics (likes, comments, views), and the key elements of a video (description, title, tags).
We used median values rather than mean values to compare italy telegram data metrics for different positions because they are more robust to outliers.
Machine learning model
We used a machine learning model to identify which features of a YouTube video are most important for ranking. Using decision tree models, we determined that title similarity to keyword ratio, view count, and video length were the most important factors for this model.
Alongside these metrics, we tracked non-engagement metrics (title length, tag similarity ratio, description length), which also showed an impact on rankings.
To compare search terms, titles, descriptions, and tags, we used the partial Levenshtein distance similarity ratio. Values of this metric range from 0 to 1, where 0 indicates that the keywords are completely different, and 1 means that the keywords match.
By exploring the top ten results for the analyzed keywords, we were able to highlight some general trends among the key factors that influence video visibility on YouTube.
Let’s take a look at the elements that shape YouTube’s unique ranking algorithm.
From our Google keyword database, we extracted 15,000 keywords that triggered a video to appear in SERP features. We then looked at the top ten YouTube search results for each of these keywords to understand how different video characteristics affect their positions in YouTube search. Based on this data, we gathered a wide range of metrics to assess which ones affect video performance.
In particular, we analyzed the performance of "how-to" keywords compared to general keywords, the positioning of videos in search engines based on their engagement metrics (likes, comments, views), and the key elements of a video (description, title, tags).
We used median values rather than mean values to compare italy telegram data metrics for different positions because they are more robust to outliers.
Machine learning model
We used a machine learning model to identify which features of a YouTube video are most important for ranking. Using decision tree models, we determined that title similarity to keyword ratio, view count, and video length were the most important factors for this model.
Alongside these metrics, we tracked non-engagement metrics (title length, tag similarity ratio, description length), which also showed an impact on rankings.
To compare search terms, titles, descriptions, and tags, we used the partial Levenshtein distance similarity ratio. Values of this metric range from 0 to 1, where 0 indicates that the keywords are completely different, and 1 means that the keywords match.
By exploring the top ten results for the analyzed keywords, we were able to highlight some general trends among the key factors that influence video visibility on YouTube.
Let’s take a look at the elements that shape YouTube’s unique ranking algorithm.