Do you often wish you knew exactly what your customers are thinking?
Is your support team constantly burned out from dealing with the ups and downs of client interaction?
You’re not alone. Businesses around the world know that their success relies on being able to interpret their customers’ thoughts and—more importantly—their feelings. But given the complexity of human communication, that can be a tall order. Customer-facing employees are frequently stuck reacting to customers in the moment rather than offering up tailored, proactive engagement.
Enter sentiment analysis. A relatively new application of technology, it promises to change the face of customer interaction forever. But how does it work? Can it really live up to its promise of making you more in tune with your customers?
How does sentiment analysis work?
Even though it sounds complex and futuristic, sentiment analysis isn’t difficult to understand. On a fundamental level, it’s a combination of two other emerging technologies: Natural Language Processing (NLP) and artificial intelligence (AI).
NLP is a subset of AI that lets machines understand human speech, the way we actually talk. So instead of only understanding a predetermined set of rigidly expressed commands, machines can comprehend a wider range of inputs that are much less structured. It’s the difference between an automated phone system telling you to “Press 1 for technical support” versus “Tell me what you’re calling about today.”
AI tools like the Watson Tone Analyzer use NLP to parse text—such as tweets, support tickets, or emails—and score it on a variety of emotions. In the case of Watson, these scores happen at both a document level (what is the overall tone of the text?) and at a sentence level so you can see exactly where those sentiments are expressed.
Because text often contains more than one emotion or tone, these scores are expressed as percentages. You can see the full range of emotions present in the text ranked according to which are most dominant. And since AI is constantly learning, the results get more accurate over time as the tool analyzes a wider body of text.
Why is sentiment analysis useful to my business?
There are lots of creative ways to use sentiment analysis. Some of the most common use cases include:
- Performing social listening by analyzing what people are saying about your brand on social media;
- Determining the content of an email inquiry; and
- Scoring customer support tickets for intelligent triage.
But why should you machines rather than people to find out what your customers are thinking and feeling?
Two words: emotional labor.
Burnout among customer-facing employees tends to be very high because their jobs demand not just intellectual or physical work but emotional work as well. Take support agents, for example. They not only need to regulate their own emotional state while working, but they also need to produce emotions in their customers. They must both remain calm themselves while speaking to an angry client and calm that angry client down, too.
Dealing with all of that can be time-consuming, frustrating, and exhausting. It means slower response times, higher staff attrition, and overall worse outcomes across the board.
With sentiment analysis, support agents don’t have to do the hard work of interpreting a customer’s communication to determine their emotional state. It can also help agents prepare better before reaching out to a customer. Sentiment analysis gives them a good idea of what to expect from a conversation with a customer before it happens, so they’re ready to provide excellent service.
On a macro scale, sentiment analysis can help you identify trends without knowledge of advanced data analytics. You can easily see how—and what—people think about your brand, the emotions they associate with your products, and where there’s room for improvement or opportunities for growth.
How can I get started with sentiment analysis? It sounds complicated.
Okay, so you want to use sentiment analysis. How do you start? It’s actually really easy to implement sentiment analysis—given the right tools.
Generally speaking, there are two big challenges with using sentiment analysis:
- You have to get the text into the sentiment analysis platform; and
- You have to act on the results of the sentiment analysis.
The easiest way to solve these challenges is to use an integration and automation platform. Intelligent Process Automation allows you to seamlessly push content from apps like ServiceNow, Salesforce, and Zendesk into a sentiment analysis tool—no coding or manual work required. You can even use automation to move text from social media outlets like Twitter.
You can also orchestrate fully automated, end-to-end workflows that use sentiment analysis to prompt meaningful actions. For example, let’s say you want to analyze the tone of customer support tickets. That’s great, but ultimately you want your support agents to be able to use that information to triage tickets more easily.
Instead of asking them to manually read, digest, and act on the sentiment analysis for every ticket, you can leverage automation to seamlessly follow up with high-priority tickets.
Let’s say you use Watson to analyze the content of a new Salesforce case. When Watson logs the dominant emotional score in Salesforce, it can kick off an automated follow-up based on the sentiment. If the dominant emotion is negative—like anger or sadness—the workflow might involve notifying a channel in Slack with the customer details so immediate follow-up can occur.
If the customer is very happy, the workflow could create a Trello card on the marketing team’s board so that they can follow up for a customer story or thank the customer via social media. This way, you can easily identify over-the-moon customers and turn them into brand ambassadors!
See It In Action: How a Luxury Retailer Uses Automated Sentiment Analysis to Promptly Respond to Customer Feedback
At Canada’s #1 luxury department store, sentiment analysis is a key part of providing a one-of-a-kind high-end shopping experience.
Like many retailers, the store asks customers to fill out a survey after each visit. The survey allows each client to rate the service, expertise, and demeanor of the staff member who assisted them during their visit.
The department store uses Workato as their integration and automation platform to orchestrate their automated sentiment analysis workflows. Survey results flow into Workday, the company’s HR hub, but they still need to be interpreted. That can slow down response times for customer feedback, so the retailer turned to automated sentiment analysis for help.
Using Workato, the company runs all customer feedback through Watson. When a customer’s feedback indicates that they’re unhappy, disappointed, or angry, Workato will automatically flag a manager via Slack or email so that the issue can be resolved as quickly as possible. The feedback is also pushed back into Workday so the company can make intelligent decisions about which employees to promote, retain, or rehire for seasonal work.