What Is Sentiment Analysis? How to Tell What Your Customers Truly Think

What is Sentiment Analysis?

Sometimes referred to as opinion mining, sentiment analysis is about perceiving and evaluating how someone feels about a certain topic and to what extent based on tell-tale factors such as their word choice, communication style, pitch and tone of voice.

Sentiment analysis is a form of text analysis or text mining, where unstructured data in surveys, reviews, social media comments, reports and articles are analysed to extract meaning.

How is Sentiment Measured?

Sentiment is measured as positive, negative or neutral and is often presented on a scale from –100 to 100 with –100 being extremely negative and 100 being extremely positive.

A Net Sentiment Score is calculating by subtracting the percentage of negative mentions from the number of positive mentions. For example, if a brand has 1000 mentions, 300 positive (30%) 250 neutral (25%) and 450 (45%) negative, the net sentiment score would be calculated as 30-45 for a total of –15.

How is Sentiment Visualised?

Sentiment over time

A sentiment over time graph can be seen as a line chart, with the number of positive, negative and neutral sentiment over time. This is a useful way to visualise how public opinion has changed over time, for example in response to a campaign, event or announcement.

Line charts are also great for visualising comparisons. For example, you can compare competitor sentiment over time, or, as in the example below, compare net sentiment against post volume to see if there is any relation.

Word cloud / term cloud

Typically, a word cloud is a graphical representation of word frequency. The more a certain word is used, the bigger and bolder it appears in the cloud.

But in a sentiment word cloud, words are sized and coloured according to their positive or negative score. Words that are mentioned positively appear green while negative words appear red.

In this example, we can see that the restaurant in question is largely perceived as clean and the service is friendly. But customers have mixed opinions about the burgers.

Word clouds are a good starting point to get an overview of the sentiment around a topic, but they don’t give you much detail. Interactive word clouds enable you to click on the words and see the full comment/review/tweet to give better context for each word.

Sentiment heatmap

In a sentiment heatmap, the colour scale indicates the intensity of sentiment, with deep red being the most negative and deep green being the most positive. A heatmap is a great way of visually representing data by topic so that you can easily hone in on problem areas.

What is Sentiment Analysis Used For?

Sentiment analysis has many practical applications in business from measuring customer response to a new product launch or marketing strategy to assessing customer service interactions at call centres, and it can be used in any industry. So, what is sentiment analysis used for? Here are some popular ways sentiment analysis is used in business:

PR monitoring

Monitor sentiment around your brand on social media to detect if there are any urgent issues and make plans to address them before they worsen.

Campaign analysis

Find out how customers are responding to your latest marketing campaign or product launch. Find out what ha

s had a positive impact on sentiment towards your brand.

Product and business development

Find out which aspects of your product or organisation are viewed positively, and which need more work and plan for necessary adjustments. Use the term cloud for clear visualisation of where your issues lie.

Brand monitoring

Quality over quantity. Brand monitoring isn’t just about counting how many times your brand is mentioned, it’s also about understanding the quality of the mentions wherever they’re found. Use sentiment analysis across review sites, news sites, forums, blogs and more.

Customer service

More advanced sentiment analysis tools can assign support queries to a topic so that you can quickly and easily direct them to the right department. With the Symanto Insights Platform, you also get further information about the personality trait and preferred communication style of the writer. Are they seeking action or information? Do they want a factual response or a more personal response?

Related Terms

Customer sentiment definition

Customer sentiment refers to the general attitude of a customer, or group of customers, towards a business, product, service, brand, etc. Customer sentiment analysis is the most common form of text analysis and is used to automatically detect, classify, and quantify the sentiment of customer feedback.

Brand sentiment definition

Brand sentiment is the collective attitude of customers towards a brand. This can be positive, negative, or neutral. Brand sentiment is often the focus of competitor analysis, as it can give insights into how customers perceive a brand in comparison to its rivals.

Product sentiment definition

Product sentiment is the reaction of customers to a product or service. As with all other types of sentiment, it is measured as positive, negative or neutral. Product sentiment can be broken down by topics or features, giving insights into which features customers like or dislike. Product development teams can use this feedback to improve future versions of the product.

Social sentiment definition

Social sentiment refers to the tone of conversations happening specifically on social media. Social sentiment analysis is limiting as it doesn’t take into account other sources of customer feedback such as surveys, reviews, and support tickets, but it is useful for marketing and PR teams who specifically want to track how customers are talking about their brand on social media.

Emotion analysis definition

Emotion analysis differs to sentiment analysis in that it goes beyond simple positive or negative language to identify the emotions being expressed. For example, a sentiment analysis tool may classify the sentence “I’m so angry I could scream!” as negative, while an emotion analysis tool would also detect the emotions of anger and frustration.

What are the Challenges of Sentiment Analysis?

Most sentiment analysis is conducted on open-ended written text, such as reviews and social media comments. However, this type of unstructured data is notoriously difficult to measure.

Language is complex and as a result, it is difficult for most technology to be able to distinguish between a positive and negative comment.

Nuanced or ambiguous language

Some sentiment analysis tools take words in isolation and appropriate positive or negative sentiment to them. But it’s easy to see how this method lacks accuracy.

Take, for example, the word “great”. On its surface, an undeniably positive word. That is until you put it before the word “disaster”. “A great disaster” or “a great shame” are decidedly negative phrases.

Natural language processing (NLP) technology is now advanced and accessible enough that, with the right tool, these issues can be overcome. However, it’s important to recognise that not all NLP technology is equally accurate in this regard.

Industry specific terminology

Similarly, the same word can have different connotations in different industries. The word “kill” is a highly negative word in the pharmaceutical industry but can be a neutral or even positive term in the gaming industry.

To counter this, some advanced text analysis tools use industry-specific models that have been trained to recognise the connotations of words in specific industries.

Multiple topics

Sentiment analysis is further complicated when more than one topic is described within the same sentence, for example, “The service was fantastic but unfortunately, the food was inedible.” Here we see that the topic “service” was highly positive, while the topic “food” was extremely negative.

There is a lot to learn from this review, but less accurate sentiment analysis tools may be unable to distinguish topics, and there’s a high chance they may tag this review as neutral overall, balancing negative words with positive words.

When choosing a sentiment analysis tool, look for one that offers topic or clause-level detection so that you can be sure you’re getting accurate insights into every aspect of the customer experience.

Accurate Sentiment Analysis Tools

Symanto’s deep learning model analyses texts in their entirety so that words are understood in context. It has a high level of accuracy when it comes to assigning sentiments towards specific topics or categories so that you can see exactly what area of your business needs attention.

The Symanto Insights Platform visualises the data into easy to understand and easy to navigate charts. These interactive charts enable you to dive deeper into the insights and find out more about what people are saying and how they are feeling.

Best of all, these insights can be available to you within a matter of minutes. The Symanto Insights Platform can scan, analyse and organise thousands of reviews not only for sentiment but also for psychographics in the time it takes to enjoy your morning cup of coffee.

Now you now what sentiment analysis is, discover the sentiment analysis capabilities of the Symanto Insights Platform for yourself by booking a free personalised demonstration, or get in touch.