Sentiment analysis for Youtube Videos & Reviews
Does your marketing strategy include uncovering brand and customer insights from Youtube videos? If not, it should. There are virtually millions of feelings and opinions about brands expressed by people on Youtube every day. Yet, when social media sentiment analysis is talked about, most people think of Twitter or Facebook, but not Youtube. This is because the idea of extracting brand insights from video seems near impossible, even though Youtube might very well be the Queen of product reviews.
Is Youtube a good source for Brand Insights?
The influence that Youtube has at various stages of a consumer’s purchase journey is undeniable. Shoppers frequently use YouTube before making purchases. Around 80 percent of shoppers claimed to have accessed a video review on the site in the early part of their shopping process. YouTube viewers watched over 219-million hours worth of review videos every year. Unlocking the brand insights buried in Youtube videos might very well be the biggest thing to happen in a very long time to social media listening, sentiment analysis, and brand insights.
Youtube has an astonishing 2-billion users globally. Although this video platform is very different from Snap, Instagram, or TikTok, it is similar to them in that ultimately it allows people to share video content with other users on the platform. People use it to view, upload, rate, share, add to playlists, report, comment on videos, and subscribe to other users’ channels. Content on Youtube is usually either user or brand generated. There is a wide variety to choose from including video clips, TV show clips, music videos, short and documentary films, audio recordings, movie trailers, live streams, and other content such as video blogging, short original videos, and educational videos. It is indisputable that Youtube is a social media platform that possesses a wide variety of brand insights buried in its video content that, if they are extracted, can help brands effectively gain more awareness and drive purchase.
Sentiment analysis is a natural language process used to identify, extract and analyze expressed opinions, feelings and emotions in text data. Text data is basically chunks of digitally produced language with a meaning intended by the author. Examples of unstructured text data are emails, social media posts, consumer reviews, customer survey answers, and business documents. Brands love to mine for opinions to bring important consumer insights to the surface that lie buried in social media posts and customer surveys. Brands such as H&M, Samsung and T-Mobile want to understand how their customers feel about their brand. Whether they are looking to establish the voice of customer, conduct market research, or monitor online brand reputation, brands use these insights to improve their products and marketing and to drive sales through more effective customer acquisition, retention and overall profitability. Basically, conducting sentiment analysis is good business.
Social media is the deepest reservoir of customer reviews, feelings, and opinions about brands. Billions (yes, billions) of people look to Instagram, SnapChat, TikTok and Youtube to either broadcast how they feel about certain brands, products or services; or watch videos of other consumers and influencers talk about a variety of brands from Walmart to Balenciaga. Sentiment analysis allows you to identify and score chunks of social media posts that express feelings, opinions or emotions as positive, neutral, or negative about any brand. Moreover, aspect based sentiment analysis allows you to be even more granular by using text analytics and named entity recognition to discover the “why” they feel the way they do about specific aspects of a brand, product or services.
What is Aspect-based sentiment analysis of Reviews?
Aspect based sentiment analysis goes one step further than typical sentiment analysis by automatically assigning sentiment to predefined categories. It involves breaking down restaurant review into smaller segments, allowing more granular and accurate insights from data. With aspect based sentiment analysis, it can be distinguished which features of a product or service offering are liked and which ones can be improved. Let’s see what happens when our reviews above become longer and more detailed:
This review needs to be analyzed at the aspect sentiment level with further aspect insights on Drinks via martinis, and Food insights are revealed through the aspects of nachos and calamari.
In typical reviews, consumers often touch on many aspects of a product or service. Complaints or praise for price, quality or ease of use can all be mentioned in one comment. Aspect based sentiment analysis determines first which categories are being mentioned and then calculates the sentiment for each of those categories. When compiled in aggregate across a large number of reviews, the strengths and weaknesses of a business’ product or services surface quickly and actionable insights become obvious instantly.
Repustate’s Deep Search for video analysis tool is able to conduct aspect based sentiment analysis on Youtube video content to deliver the most granular, useful and actionable brand insights. In addition, it also conducts Named Entity Recognition (NER). NER is an advanced natural language processing technique that uses automation to identify named entities in your Youtube videos, and classifies them into categories that are predetermined. NER can be used to classify company names, geo-locations, things, and names of people that are mentioned in your videos and present them to you in a way that brings your videos’ insights to the surface. These insights can be used to improve your brand’s marketing efforts, products, customer experience, or customer service.
How does Repustate`s video analysis tool perform sentiment analysis on Youtube?
Repustate’s Deep Search for video analysis tool extracts brand insights and performs sentiment analysis in these 3 simple steps:
- Convert video to text through the use of speech-to-text models
a. These models are unique to each language and are built using neural networks
b. This yields a transcript of the speech along with the timestamps for each word
2. Index the text into Semantic Search for Video section by section. Video can be broken up into smaller sections based on longer pauses or changes in the speaker. By analyzing it section by section, we get enough context to disambiguate entities, but not too much content as to lose the ability to associate entities with granular enough timestamps.
3. We now have timestamps of video associated with entities, themes, and topics found in the text, along with the associated metadata. Performing Deep Search will now yield all matching sections, which in turn refer to snippets in the source video.
What does Unbox therapy think of the new PS5?
One of my favorite Youtube channels is Unbox Therapy that consists of new technology unboxings like smartphones, laptops and gaming consoles often before they are released to the public. It’s produced by Canadian Lewis George Hilsenteger, and with 17.6-million subscribers, Unbox Therapy has an incredible influence on consumer tech purchases.
In a recent video, Hilsenteger reviews and compares two of the biggest Christmas 2020 gaming console releases — Sony’s PS5 vs. Microsoft’s XBOX Series X.
https://www.youtube.com/watch?v=Jq-ODza3Kpc&t=624s
This video was posted on November 6 and already has an amazing 5,750,761 views. So being a Canadian company dominated by tech heads, we figured we’d put this video through Repustate’s Deep Search for video analysis tool and pulled our some insights super quickly and accurately.
At between the timestamps 5:54 and 6:27, Hilsenteger says:
Here the theme is styling in reference to the two entities PS5 and the XBOX Series X. He says that the PS5’s styling feels “complex” and “newer” and will “command a little bit more attention”. The newness comes from the product’s aspects “tapering”, “shape”, “motif” and “lighting”. On the other hand, the XBOX Series X appears “simple” and logically less new than its competitor.
On the topic of controllers at between 7:07 and 7:14 the Canadian tech influencer says:
So when it comes to controllers, the PS5 wins again as Hilsenteger goes on to point out Sony’s superior product design and newer styling.
These are all incredible brand insights that Repustate’s Deep Search for video analysis tool indexed and analysed in mere seconds. Imagine how much time you would save, and how much invaluable product intelligence you could gather, if you did this for hundreds or thousands of Youtube videos that talk about your brand? That’s where the real power and intelligence of Repustate’s Deep Search for video analysis tool lies.
Originally published at https://www.repustate.com on December 8, 2020.
