Do Bots Speak Emoji?

Businesses always want to know more about what their customers are thinking and feeling. Organizations often use survey tools to gauge customer opinions, but the insights can be limited by low response rates and the quality of the survey questions.

AI-powered solutions that measure customer sentiment can help businesses overcome these limitations and provide actionable information that drives business results. It’s another example of how artificial intelligence is transforming contact centres.

In these days of social media, people let it all hang out. Emotions and opinions that in the past would have been kept private are now on the internet for all the world to see.

People have been emboldened to make snarky comments on Twitter to complete strangers, post incendiary product reviews, and share videos of their toddler’s tantrums. And we all have friends who overshare on Facebook, right?

While public emoting is a treasure trove of consumer information to organizations that have the right tools to mine it.

Businesses that use software capable of customer sentiment analysis can monitor contact centre interactions, social media posts, product reviews, blog posts, and more.

They can then use that information to understand brand perceptions, identify product flaws or gaps, perform competitive analysis, and build a more holistic voice of the customer (VoC).

What Is Customer Sentiment?

Customer sentiment includes people’s attitudes, emotions, and opinions. In other words, it’s subjective information that can be difficult to discern. Have you ever received an email where you thought the sender was angry with you, but they weren’t?

And how about text messages from teenagers who use slang and no punctuation? Fortunately, some smart person invented emojis to help us express emotions in our digital communications.

Customer sentiment is invaluable information for companies to have, but measuring it is complex because of people’s different communication styles.

Fortunately, the quality of artificial intelligence is advancing, and organizations, including call centres, now have access to tools that can quickly comb through huge amounts of data to perform sentiment analysis in real time.

What Is Sentiment Analysis and How Does the Technology Work?

Sentiment analysis, also called opinion mining, is the process of assessing written and verbal customer inputs to determine what people are thinking and feeling about a brand, products and services, customer service interactions, and other aspects of the business.

This customer input is typically scored as positive, negative, or neutral.

Humans are still better than machines at detecting sentiment, but they’re slower at it and it would take an army of employees to stay on top of all of a brand’s online mentions or to review every single customer service interaction.

And this army wouldn’t produce perfect results because of the complexity of human speech we’ve already discussed.

Sentiment analytics software, on the other hand, can speed through analysis in real time. And the software uses rules, so the results are more consistent than what a team of multiple humans can produce.

These AI-powered solutions use natural language processing (NLP) to understand human language and machine learning to get smarter with additional data consumption.

These tools need to be trained – by consuming large amounts of data – to understand nuances of human conversation such as intent and context.

The best off-the-shelf solutions come already trained. For example, the NICE Enlighten AI platform has been trained by consuming over two billion human interactions.

Sentiment analysis uses keywords, context, and characteristics of voice conversations to determine if customer sentiment is positive, negative, or neutral.

For example, if, during a customer service chat session, a customer says “I’m disappointed with your product,” the analytics tool would score the interaction as negative.

But if the customer says, “I’m disappointed with your competitor’s product,” hopefully the analytics tool would score that as neutral or maybe even positive.

And, yes, sentiment analytics tools can interpret emojis.

Interpreting voice interactions is a little different. Sentiment analytics solutions still look at keywords and context, but it has additional data to factor in – speech characteristics. Volume, pitch, and pace of speech can indicate if someone is relaxed or agitated.

And long, disapproving silences after the other person says something can speak louder than words. Additionally, if the customer and agent repeatedly interrupt each other, it can indicate they’re both unhappy.

Customer sentiment can be calculated in the aggregate, to give organizations an overall view of how customers are feeling, and drilled down to individual customers to identify people who need additional attention.

When used in call centres, customer sentiment can be calculated at the agent level to identify training opportunities and reward agents who are performing like champions.

Customer Sentiment in the Call Centre

As you can imagine, there are plenty of ways sentiment analysis can be applied in call centres. Conversations between customers and agents are rich sources of customer sentiment that can be used to identify issues, guide agents during interactions, and make contact routing smarter.

The full potential of this customer data was previously unrealized because it just wasn’t practical for humans to sort through it all and make sense of it. Artificial intelligence has changed the game and businesses can now take full advantage of this valuable customer information.

Below are some examples of call centre tools that leverage the power of sentiment analytics.

Interaction Analytics

Interaction analytics software is capable of assessing all contacts from all channels to identify emerging problems, root causes, contact drivers, compliance issues, and more.

It can also use sentiment analysis to determine what customers are thinking and feeling. Interaction analytics tools provide meaningful customer insights that lead to better decisions and better results on key metrics like first contact resolution.

Additionally, customer sentiment insights provide a more accurate and comprehensive view of customer satisfaction.

Real-Time Interaction Guidance

Real-time interaction guidance solutions coach agents during every interaction on hard-to-train soft skills.

This AI-powered tool leverages sentiment analysis to identify how customers are feeling and can alert agents to, for example, show more empathy, use more active listening skills, or slow down their rate of speech.

This is a significant improvement over traditional interaction coaching, in which agents receive feedback on a small percentage of contacts several days after the fact. Interaction guidance allows agents to improve conversations while they’re happening, which should lead to more customer satisfaction.

AI-Enabled Interaction Routing

When artificial intelligence is applied to call routing, it can make one of the oldest call centre processes seem fresh again. AI-enabled interaction routing factors in a multitude of variables, such as customer personality and preferences, to optimize customer–agent matching.

It can also assess sentiment of incoming interactions and route accordingly. For example, if a customer sends an email that’s angry, that email can be routed to an agent who is specially trained at de-escalating tense conversations.

What Actions Should Businesses Take Based on Customer Sentiment?

Knowing how your customers are feeling is invaluable information that businesses can use to improve CX, training, product quality, and many other critical drivers of business results. Here a few ways to leverage sentiment analysis.

Close the Loop With Dissatisfied Customers

Perhaps your first priority should be closing the loop with at-risk customers. Knowing that customers are unhappy gives you a chance to save relationships before customers walk out the door. There are several ways to close the loop.

For unique issues, it may be appropriate to have supervisors call customers and work through problem resolutions one-on-one. For more general widespread issues, such as a product flaw, a group email might do the trick.

Timely, effective interventions with disgruntled customers can repair relationships and make them even stronger.

Improve the Customer Journey

Customer sentiment analysis enables organizations to identify root causes of issues, including friction at specific touchpoints along the customer journey.

For example, AI analytics tools can determine if customers are particularly unhappy with a marketing promotion, retail experiences or website functionality.

Businesses can use this information to enhance all touchpoints, so customers have consistent experiences regardless of the path they take. Following implementation of improvements, organizations can monitor customer sentiment to see if the changes made a difference.

Develop and Reward Agents

When sentiment analysis is used in contact centres, it provides additional insights about agent performance.

That’s because customer sentiment scores can be calculated at the agent level and used to pinpoint needed training as well as profile characteristics of high performing agents. Some contact centres even tie a portion of agent compensation to customer sentiment scores.

But organizations that use sentiment to evaluate their agents need to proceed with caution – many times customer anger and frustration is directed at the issue, not the agent.

Improve Products

Customer sentiment can be calculated by topic, allowing businesses to identify hot spots. When those hot spots are product-related, product development teams can use that input to fix or enhance products.

Additionally, this customer feedback can help identify gaps in the product portfolio that can be filled with new products. Addressing customer needs through new or improved products will lead to better CX and more revenue.

Address Problems

The ability to identify hot spots also enables organizations to proactively identify and manage new problems.

For example, if customer sentiment is more negative than usual about billing, contact centres can drill down on the data and see if there’s a new problem related to billing that needs to be addressed.

Catching problems early, before they become fires, allows businesses to quickly fix issues, proactively communicate with customers, and avoid related contact volume.

Limitations of Sentiment Analysis

As you can see, sentiment analysis has the potential to transform customer relationships, but it has some limitations. It’s not all sunshine.

Human speech is complex and it can be hard for machines to understand nuances like sarcasm and humour.

For example, if a customer service agent tells a caller that their flight has been cancelled and the caller responds with “Wonderful! That’s just great!” sentiment analytics might score that interaction as positive, while most humans would realize the caller is using sarcasm to express their anger.

Sentiment analysis can also struggle when part of a sentence negates another part. For example, for the phrase, “Well, I didn’t hate the chocolate,” the sentiment analytics software might zero in on the word “hate” and give this statement a negative score when it should probably be scored as neutral.

These limitations shouldn’t prevent you from measuring customer sentiment in your call centre, but it’s important to know that sentiment analysis isn’t perfect – yet.

Customer sentiment analysis is one of many AI-powered solutions that’s transforming the way contact centres operate.

This blog post has been re-published by kind permission of NICE CXone – View the Original Article

For more information about NICE CXone - visit the NICE CXone Website

About NICE CXone

NICE CXone combines best-in-class Omnichannel Routing, Workforce Engagement, Analytics, Automation and Artificial Intelligence on an Open Cloud Foundation.

Find out more about NICE CXone

Call Centre Helper is not responsible for the content of these guest blog posts. The opinions expressed in this article are those of the author, and do not necessarily reflect those of Call Centre Helper.

Author: NICE CXone