Your Genesys Blog Subscription has been confirmed!
Please add [email protected] to your safe sender list to ensure you receive the weekly blog notifications.
Don't Show This Again.
When Doris Day sang the song Sentimental Journey back in 1944, it’s unlikely she could have imagined applying sentiment analysis to speech analytics in contact centers. However, the desired outcome of the song and sentiment analysis are one and the same: to put one’s mind at ease.
To improve their teams’ performance, contact center managers want to answer some questions, including:
There are several ways to find the answers to these questions: listening in on the call either in real-time or after the call, reading a transcript of the call to see if it was perfect, sending out a survey to the customer after the call happened, or even asking the employee to assess their own performance. But all of these methods have their flaws.
Listening to recordings of interactions can be time-consuming. Reading over transcripts, while potentially faster, is as well. Post-call surveys give you an idea of customer sentiment and Net Promoter Score, but they typically assess the entire system rather than any individual agent in the interaction.
Self-assessments also pose challenges. At one extreme, modest agents might underrate their performance while more confident agents might think their customer interactions went better than they actually did. Even if accurate, these assessment tools don’t scale well, nor do they provide quick feedback on what’s happening on the ground — and in the moment.
Sentiment analysis looks over text and applies sentiment markers. Was a phrase spoken by a customer positive, negative, or neutral? This data can then be analyzed in different ways to give a sense of how the individual agent or contact center is performing.
Along with its siblings, entity, and intent recognition as well as topic spotting (among others), sentiment analysis is part of the natural language understanding family. These services rely on artificial intelligence (AI)-derived methods for looking at training data and then predicting whether a particular language will match certain criteria, such as being positive or negative in sentiment.
The main advantage of applying sentiment analysis to all interactions is that it scales quality management. Every interaction can be assessed versus the few that a manager has time to listen in on.
In a contact center, sentiment analysis sits toward the top of the speech and text analysis hierarchy. At the base, you need to record interactions. Then you need to convert these recordings into text through speech recognition. Once this is complete, you can apply different analyses, such as topic spotting, phrase detection, and sentiment analysis. You can also apply other analyses to the recording, such as acoustic analysis, speaker recognition, or emotion detection based on tone, pitch, and other voice qualities.
Contact center managers can use sentiment interaction in a few ways. First, if marked on a transcript, they can quickly see where a customer expressed positive and negative sentiments. This allows them to narrow down potential issues or quickly find examples to highlight excellent performance.
Second, managers can see whether interactions are trending in the right direction over the course of a call. For example, it can tell if a call started off with poor sentiment and then moved to a more positive sentiment toward the end of the call. If that happens, it might be an example of how an agent is effectively resolving a customer issue.
Sentiment analysis also has the potential to automatically surface interactions for further quality assurance by supervisors, instead of relying on supervisors to search for these manually.
One of the powerful outcomes of applying sentiment analysis to all interactions are the insights it offers a contact center manager. Companies can see whether their interactions lead to better trends during the call, e.g., from negative to positive. They can also see whether, over time, customers are engaging with the contact center from a more positive starting point. This could signal improvements in product performance. Negative sentiment might also serve as an early warning for product or service defects that you can act on immediately.
Sentiment analysis gives contact center managers more tools to scale quality management. And that means they can refocus efforts on helping their team become better skilled to offer superior customer experiences. For more information, check out the Frost and Sullivan whitepaper about how to engage your team for the best CX.
Subscribe to our free newsletter and get the Genesys blog updates in your inbox.