How does sentiment analysis help my video research?

Video is by far the most powerful way to gather human feedback and see how your customers really feel about your brand, products, and services. However, video market research has had a bit of a bad reputation over the years, with many researchers being reluctant to use it because they see video as cumbersome to record and even more complicated to analyze. But with today’s end-to-end video research tools, there’s no need to spend hours searching for answers or carefully noting down time-codes, consistent topics and annotations by hand. Instead, advanced, automated analytics such as theme coding and sentiment analysis are empowering researchers, allowing them to quickly understand shared sentiments and consistent themes across hundreds of videos, or hours of content, at a time.

What does that mean for you? That you can access the depth of information delivered by video and use it to make informed, customer-centric decisions faster than ever before.

Sound good? Read on to find out more…

What is automated sentiment analysis?

Sentiment analysis is a video analytics tool available within Voxpopme that is designed to bring you closer to your customers’ feelings towards your products services, adverts and more. It works by reviewing every sentence of each video uploaded, determining whether it is positive, negative or neutral and giving it an associated score – so you can understand the sentiment behind every single sentence, in every single video response and understand the subtle nuances of every comment.

It auto-categorizes content by respondents’ feelings and allows you to effortlessly explore sentiment across wider themes too. It’s the ideal blend of human and machine analysis, removing the margin for human error in the categorization stage but still allowing you to pull out the video snippets that are most informative within a category. Essentially, it means you can build an understanding of sentiment without the human biases that are often present in manual analysis, whilst at the same time vastly increasing speed, scalability, and accuracy. Basically, it allows you to spend less time searching and more time telling stories that resonate – bringing you closer to what your customers think than ever before.

How it works

Automated analytics can be added to video of any length using an end-to-end video insight platform like Voxpopme, whether it’s short format content recorded by consumers or existing long-form content (such as focus groups or IDIs) that you’d like to upload to reduce the analysis burden of traditional video. Once uploaded, the analysis essentials like human transcription, time-coding, quality checking and more will quickly take place to allow for the more advanced, automated analytics like sentiment coding to kick in.

At Voxpopme, sentiment analysis is powered by IBM Watson and uses machine learning and natural language processing to identify the underlying sentiment within each individual sentence of your videos. Once a video is uploaded into the platform, it processes the transcribed text and returns a polarity on every sentence, determining whether it is positive, negative or neutral. It then scores each sentence within the transcript between -1 (negative) and +1 (positive) with 95% accuracy.

Sentence level sentiment means you can identify precisely how respondents answer each of your questions, associating attitudes with every sentence of their responses to give you a thorough understanding of how they really react. It then displays the percentage breakdown of sentiment via interactive bar charts, showing the percentage of positive, neutral and negative responses for a quick and easy overview of the sentiment breakdown:

Not only that, but there’s also a color overlay on each sentence in the video transcript too, so you can instantly see whether the overall feel of the video is positive, negative or neutral as well as identifying the sentiment in individual sentences. In addition, by hovering over each sentence you can also see the sentiment score- making it easier than ever before to understand and analyze customer videos and feedback.Sentence Level Sentiment Analysis

If that wasn’t enough, a deeper level of analysis is also available when using sentiment analysis by utilizing automated theme coding such as Theme Explorer to demonstrate the sentiment breakdown of each theme discussed within your videos. The sentiment chart for each theme will give you an instant indication of how respondents are discussing each theme, as well as looking at the percentage breakdown of sentiment for each theme.

That means that instead of just understanding what respondents are expressing, for example, ‘63 respondents mentioned quality’, you can now quickly identify the sentiment behind what your respondents are saying. For example, ‘72% of respondents spoke positively when talking about the quality of our brand’ – providing you with a clear, holistic picture of what your customers are thinking and giving you all the information you need to create compelling stories that are worth telling.

Sentiment analysis overview

What are the benefits?

By categorizing content according to your respondents’ true feelings, you can build a deeper understanding of large volumes of video, in less time, to establish tangible insights – which means you can understand what your customers are thinking better than ever before and drive real business outcomes. And you can do all this without having to trawl through endless video responses or surveys to do so – just think of the amount of human analysis time saved!

Not only that, but because it’s so easy to use, you can stop intensively searching your footage on the lookout for answers – because they’re right in front of you. Which means you no longer need to spend your time struggling with clunky video analysis – you can now easily create powerful showreels telling customer-centric stories to get your customers’ voices heard by the people that matter.

To sum up, sentiment analysis means you can determine exactly how your customers answer your questions by associating attitudes with every single sentence of their responses. By creating a sortable and searchable picture of how your customers answer each question, you can get to the bottom of what they really think so you can quickly find the insights you need to deliver a much bigger, clearer picture and make informed decisions, fast.

Like what you see? There's more great content where that came from.

Check out white papers