Emotion Analytics and Sentiment Analysis are easily misunderstood. Sure, they both seek to understand nuances in customer data, and they each have a role to play in Human Experience strategy. But let’s get one thing straight: They are not the same thing and there are three key reasons why this is the case, particularly when it comes to improving Customer Experience.
What is Sentiment Analysis?
Sentiment Analysis is a coarse-grained approach that tells you if something is positive, negative or neutral. This does not allow for an understanding of how you might actually act upon this information. The problem with data categorised as ‘neutral’ is that there is high potential to miss valuable insight. Sentiment analysis works a bit like poker in this sense – if the keyword used to categorise posts as ‘positive’ or ‘negative’ doesn’t occur, the system will essentially ‘fold’ and classify the text as neutral. Additionally, more advanced techniques suffer from the problem of ‘overfitting’. You can learn from labelled data, but this means that it has been trained for a very narrow use case. This makes the system incredibly fragile and hard to scale. ‘Without knowing why your customers feel the way they do – or having an accurate understanding of how they’re actually feeling, beyond just “positive” or “negative” – it’s impossible to identify ways to improve’ (CX Magazine, 2017).
What is Emotion Analysis?
Emotion analysis, by contrast, entails a fine-grained, deep dive into the themes associated with each emotion and allows businesses to discover which emotions impact Human Experience. Emotions, by their very nature, are both continuous and subject to constant change and thus Artificial Intelligence (AI) is required to understand and interpret them. Emotion AI pinpoints these changes based on the intensity of the emotions people are feeling. For example, instead of blankly revealing that 10% of customers respond negatively to a certain brand, emotion analysis will reveal three key pieces of information. First, it will reveal what the emotion is. Second, it will show how intense this emotion is, and third, it will tell you what the emotion being expressed relates to.
Let’s take a look at 3 reasons why you should analyse emotion instead of sentiment when it comes to understanding CX and human experience.
1. Sentiment analysis oversimplifies the data:
Dividing things into positive and negative sentiment isn’t meaningful, as it doesn’t provide the full picture. For decision-makers, sentiment does not help to prioritise the importance of data. Instead, it places data into three boxes labelled ‘positive’, ‘negative’ and ‘neutral’. Human emotion is so incredibly complex that it required much more than traditional sentiment analysis to understand it. Emotion analysis can break down this 3 point scale into key emotions such as joy, anger and trust. This provides a more nuanced and accurate view of emotion than the traditional “positive-neutral-negative” options in sentiment analysis.
2. Sentiment can’t answer business questions in the way emotion analysis can:
Simply put, if sentiment analysis provides the ‘what’, then emotion analysis goes far beyond this to explain the ‘why’.
The key difference is that ‘emotions capture both polarity and intensity, where sentiment is limited to one dimension’ (Walker, 2016). If you ask a sentiment analysis engine to determine the biggest emotional driver for customer churn, at best you will get a percentage of ‘positive’, ‘negative’ and ‘neutral’ mentions, along with a topic wheel. This is not fit for purpose and certainly not something that could be presented at board level. Emotion analysis provides an exploratory approach – identifying the contributing factors to each emotional index so you not only discover the root cause but are also able to prioritise the decisions you need to make.
3. Approaches need to match real-world use cases:
The shift towards emotion analysis has been driven by a number of market factors. In part, the underlying industry response to sentiment analysis is: We can do better than this. Secondly, the rise of Customer Experience, and how this disrupts how brands engage with customers, also plays a role. When customer loyalty is on the line, you need to understand what elements of the Customer Experience are driving customer disgust, so immediate action can be taken. Forrester, in their top trends for 2017, state that firms will begin to quantify and better harness the power of customer emotion to drive affinity and spend.
Leading academics, top marketers and market research companies are calling for new forms of emotion analytics that effectively render sentiment analysis obsolete. Professor Bing Liu of the University of Chicago has urged businesses to seek alternatives to “the problem of sentiment analysis” (Bing Liu, 2012). Forrester’s 2019 Customer Experience Index revealed that “emotion plays a critical role in differentiating brands and has a bigger impact on brand loyalty than effectiveness or ease of use, regardless of industry”.
