If you (or your platform) are still only looking at sentiment to understand customers or employees, it’s time to drill down into the deeper detail.
Thanks to AI and machine learning, the science of text analysis is becoming increasingly intuitive when it comes to capturing and analysing qualitative feedback. However, like many other innovative technologies, the way that text analysis is actually being applied is advancing much more slowly. If you (or your platform) are still looking at sentiment, it’s time to drill down into the deeper detail and get an understanding of emotion: what your customers and employees are thinking, and how those thoughts and reactions to their experiences with your brand and are making them feel and behave.
Sentiment and emotion in insight – what’s the difference?
Sentiment is usually measured on a linear scale (from positive to negative) with a varying number of stages points along the way, usually three or four. Identifying that something you are doing is making people feel positive, negative or even neutral is certainly important.
Natural language processing (NLP, not to be confused with neuro-linguistic programming) applies context to recognise not just keywords, but also the context in which they have been used, plus any negations, conditions, qualifiers or amplifications associated with them. In short, NLP enables systems to interpret complex sentences more accurately. Instead of simply flagging the words ‘pleased’, ‘happy’ and ‘simple’, for example, it can recognise the negative positioning of comments such as “I would have been pleased”, “I found the whole thing over-simple”, “Apple has never been this bad”.
However, once you move beyond NPS and add emotion to the reporting mix, things get more interesting. Three different people recognised as expressing negative sentiments about the service they received may actually be feeling three completely different emotions, or blends of emotions, and each at different levels. For example, a late delivery could result in frustration, anger or bitter disappointment. All of which are certainly ‘not happy’, but also not the same thing or necessarily caused by the same reasons. It’s important to remember that emotions are not a subset of sentiments, as many that can be expressed as either positive or negative – consider anticipation, embarrassment, worry or surprise.
Why include emotion in text analytics?
Brand engagement is based on the emotional responses of the people you interact and transact with. It could be argued that positive and negative scores are all you need to know, but that would hardly be doing justice to all the juicy qualitative feedback that’s now available to you through text analytics. Ignoring the emotion would be like getting off the train a couple of stations short of your destination – you’re not completing the insight journey.
How can emotion be measured?
Emotics uses Plutchik’s Wheel of Emotions, which is one of the most consistently recognised and referenced models (it’s nearly as famous as Maslow’s Hierarchy of Needs, but not quite). The value of Plutchik’s Wheel of Emotions is that it can be used to identify emotional intensity (low, medium, high) and then quantify the actions that people will take. For example, you can quantify the themes that move from annoyance (low), anger (medium) to high (rage). Adoreboard’s Emotion AI places the 24 emotions into 8 Emotional Indexes and uses advanced theme detection to pinpoint specific issues to specific emotional intensities. In this way, you can quantify specific issues to action.
Framework in action: Decision Ready Insights
This framework is used to measure emotional intensity about your brand, and index and benchmark how well it is performing. Crucially, that information can then be used to undertake root cause analyses that will show you exactly why people feel the way they do. This is the significant advantage of emotions analytics over and above systems that are based purely on keyword approaches, and it’s why we’re always on about the value of the decision-ready insights that result, including SWOT analysis, competitor benchmarking, customer journey mapping and measuring brand empathy. If sentiment analysis can tell you what has happened in a binary positive, negative or neutral, emotion AI upgrades this to insight and action by telling you why.
Using emotional insight to drive innovation, improvement and profitability
One of our clients used Emotics to map their customer journey and discover the pain points that were making customers hot under the collar. By taking positive action to fix the issues identified, the company was able to increase brand advocates by 500% and reduce detractors by a thumping 74%. Quantifying the emotional reaction to their messaging enabled another client to improve the way in which they communicated with their customers to improve engagement. An online organisation also managed to optimise their cost per click by 50%, significantly reduce their bounce rate and up their brand consideration by 2%, by having a clearer and much more detailed understanding of customer reactions and generating enhanced content in response.
So bear in mind that not all text analytics systems are created equal: make sure you’re capturing and analysing emotional insight to get to the heart of human experience.