YouTube Insights From Video Analysis - MAC Cosmetics

YouTube video analysis can boost your brand intelligence and marketing campaigns by giving you deep insights from video content - something you can’t get from social listening tools that harness insights only from text.

Most companies obtain social insights from quantitative metrics that tell you the number of likes, dislikes, video views, shares, etc. Others, by analyzing text data comprising comments. However, AI-driven video analysis lets you derive YouTube insights not only from key performance indicators and comments but the videos themselves.

The capability to capture visual and audio content is critical when it comes to extracting intelligence from video-heavy social platforms. Thus, YouTube video analysis is by far the most effective method in measuring campaign success and conducting voice of the customer analysis.

In this article, we show you how precisely you can get this advantage. For the project, we have chosen a MAC cosmetics video from award-winning makeup artist, Hindash @hindash.

Repustate IQ YouTube Video Analysis In Action

To showcase the insights generated by Repustate IQ for YouTube video analysis, we chose makeup artist Hindash’s viral video for MAC Cosmetics. Hindash was the chosen artist from the Middle East by MAC Cosmetics and was given the opportunity to create a new lipstick for their line.

His showcase video went viral across social media. For the purposes of this example, we chose the video on YouTube. We ran it through our video AI-powered sentiment analysis platform, Repustate IQ, and below are the results.

Repustate IQ collected all the comments and hashtags from the video and analyzed them along with the video itself for sentiment. It then generated the insights on its sentiment analysis dashboard. Let’s take a look at some key discoveries.

1. Comments mentioned over time with sentiment

Repustate IQ quickly gives you the sentiment analysis of all the comments mentioned in the video, along with the sentiment trend. We can clearly see that the sentiment is highly positive and at a peak throughout the period between March 2020, when the video was released, until the end of the year. Comments mentioned over time with sentiment YouTube Video Analytics

2. Sentiment analysis over time

You can see the distribution of sentiment over time in the form of bar graphs as well as a pie chart through simple color codes. In these YouTube insights, you can see that there was positive sentiment throughout early 2020 to mid-2021, and then suddenly all sentiment was neutral between August 2021 to September 2121.

Scores like this prompt you to examine the reason for this trend. Was it a campaign that was running during the time or a pause in the marketing plan? Sentiment analysis over time in YouTube Video Analytics

3. Aspect trend

YouTube video analysis of the video showed the trend in which aspects were occurring over a period of time. In the below screen-grab we can see the aspect of “appearance” peaked twice - once between Sep 2021 to Nov 2021, dipping, and then again rising during Jan ‘21 and Feb ‘21.

Similarly, people talked about staff in March 2020, not much thereafter, until they started talking about staff again during the two peak periods depicted in the graph.

Aspect trend YouTube Video Analytics

4. Aspect by sentiment

These YouTube insights tell us how negative and positive the sentiment for each aspect is. Interestingly, the high negative sentiment linked to the aspect of availability shows that the lipstick was so popular that the company miscalculated its popularity and thereby, the distribution of the product, which upset many people. Aspect by sentiment analysis in YouTube Video Analytics

5. Aspect-Topic Breakdown

You can clearly see the sentiment breakdown for each aspect and topic.

Aspect topic breakdown in YouTube Video Analytics

By clicking on the negative or positive scores, you can see the comments that made up the score for that particular topic and aspect.

Aspect topic breakdown with sentiment in YouTube Video Analytics

Check out other insights derived from aspect-based sentiment analysis.

6. Hashtag Analysis #macxhindash

Next, we analyze sentiment related to all the videos and comments that use the hashtag #macxhindash. We see that the hashtags have a high positive score.

Hashtag analytics in YouTube Video Analytics

7. Sentiment by Volume and Source

We can see the distribution of sentiment based on data sources as well as where the chunk of the data came from.

Sentiment by Volume and Source in YouTube Video Analytics

8. Hashtag Aspect Aentiment

We see YouTube video analysis data with sentiment based on aspects gathered from the hashtag. For example, in the below screenshot we see that customer experience has been mostly positive, among other things.

Hashtag aspect sentiment YouTube Video Analytics

9. Languages

The tool gives you the breakdown of sentiment based on the languages in which the comments are. You can see that the algorithm has recognized 5 different languages in the comments.

Multilingual sentiment analysis in YouTube Video Analytics

10. Common Words and Phrases

These YouTube insights show you all common words and phrases that have occurred in the videos and comments pertaining to the MAC hashtag.

Common Words and Phrases in YouTube Video Analytics

11. Total Sentiment score

You get the sentiment score by the highest or lowest score, and by time period as seen in the screen-grab below.

Total sentiment score in YouTube Video Analytics

Learn about the other insights derived from sentiment analysis software.

Take a quick tour of Repustate's Video analysis solution.

From Which Sources Does A YouTube Video Analytics Tool Gather Data?

A YouTube video analytics platform collects data from three sources - the YouTube dashboard, comments and hashtags, and the videos themselves.

Sources For Collecting Data for YouTube Video Analytics

  1. Dashboard Analytics - The YouTube dashboard analytics gives you metrics on key performance indicators like the total number of comments, shares, likes, dislikes, etc.
  2. Comments and Hashtags - This is the source of all the text data comprising comments as well as hashtags related to your video or videos.
  3. Videos - This data source consists of the YouTube video itself. YouTube insights from this source come from either one video, or all the videos in a particular channel, or those related to a hashtag or keyword.

How Does YouTube Channel Analytics Work Within Your Profile?

The YouTube Analytics API allows you to create your own custom dashboard from the key performance indicator data you want from your YouTube channel. These are -

  1. Likes count - Number of times your video was liked
  2. Dislikes count - Total number of times the dislike button was hit
  3. Shares count - How often your video was shared
  4. Views count - How many times your YouTube video was watched
  5. Comments count - The number of comments your video has gained under it. This includes reply threads
  6. View duration - The number of seconds or minutes your video was watched
  7. Subscriber count - The total number of subscribers your channel has
  8. New subscriber count - How many new subscribers you have gained

How Do We Get YouTube Insights From Comments & Hashtag Analysis?

Insights from YouTube comments and hashtags-related data are a very crucial source of business intelligence for branding and marketing efforts. Here are the three steps in which YouTube insights from comments and hashtags is gathered.

Step 1: Data Preparation (comments & hashtags)

First, you need to pre-process the data by cleaning it. You do so by removing all redundant words, hyperlinks, and non-text data. You can either use the YouTube API to get this data or a comments scraper - yt-comment-scraper - npm. All this data must be in a .csv file so the YouTube insights platform can ingest it for processing.

Step 2: Processing the data for sentiment

In this step of YouTube video analysis for comments, all the cleaned and prepped data runs through a sentiment analysis API. It is processed for sentiment after natural language processing tasks have extracted relevant information about key aspects, themes, and features occurring in the data. Named entity recognition (NER) ensures that any important named data like a place, brand, important person, currency, etc. is extracted for sentiment analysis as well.

Step 3: YouTube Insights Visualization

The tool showcases all the insights it has gained after processing the data, in the form of charts, graphs, color codes, and other visualizations through our sentiment analysis dashboard. You can see how positive, negative, or neutral the feelings are towards your video, and by extent, your brand. You can also see sentiment for aspects and features that the tool extracted from your video. For example, color, room, fit, price, convenience, and such, are aspects.

Learn in greater detail about YouTube comments analysis.

Take a quick tour of Repustate's Video analysis tool.

How We Do YouTube Video Analysis?

A machine learning platform that has video content analysis capability can harness intelligence from video and audio data. This lets you get even deeper YouTube insights by analyzing not only comments and hashtags but the videos themselves. These could be multiple videos or any particular one. There are six stages in which YouTube video analysis happens.

YouTube Video Analytics

Stage 1. Audio transcription

All audio in the video is transcribed. This works well on all audio formats including podcasts.

Stage 2. Caption overlay extraction

The video is segmented into frames. The tool then scans the frames and extracts text in any caption overlays it finds.

Stage 3. Image & logo recognition

In this stage, the tool searches and extracts any graphics like logos, identifiable images of locations, places, etc, in the video and extracts them. Knowledge graphs and semantic analysis help the tool to identify logo in the images and piece together the puzzle. For example, with the images in the background, it will tell you the video was shot at a Subway takeaway counter, near the Yankee Stadium, and therefore, in New York.

Stage 4. Text analysis of comments

All text in comments is extracted and added to the data collected from the transcription from Stage 1. This is now processed and analyzed by the Repustate Text analytics API.

Stage 5. Sentiment & semantic analysis

All the data is now processed for sentiment analysis in depth, thus giving you not only the overall sentiment discovered for the video but also aspects of the video.

Stage 6.YouTube Insights Visualisation

All the YouTube video analysis insights are now available on a sentiment analysis dashboard in the form of charts and graphs.

The same concept is applied when searching for insights from other video-based platforms such as TikTok consumer insights or from Instagram or Facebook videos.

Read more for more in-depth information on YouTube sentiment analysis.

Summary

These are just a few of the YouTube video analysis insights that a sentiment mining platform powered by video content analysis like Repustate IQ can give. The platform gives you the ability to leverage all the data that you have access to on YouTube, so that you can improve brand experience, gain competitor insights, and business intelligence not only from comments but the video data itself. Highly customizable, you can add or remove aspects, add sentiments, and even set alerts for keywords, hashtags, and spikes in mentions.

See Repustate IQ in action and find out all the other granular insights it can give.

Book a free demo today and explore the platform!

Talk to our experts for more information