TDWI Upside - Where Data Means Business

Q&A: How Speech Analytics Detects Customer Sentiment

Speech analytics is increasingly being used to help companies store and analyze customer conversations for data-driven decision making.

Customer call centers give companies a rare chance to hear directly from customers, in conversations that can be rich with data. Recognizing that, companies are increasingly using software that captures and analyzes customer speech. Calabrio is one such company; in 2015, Gartner named it a leader in customer engagement center workforce optimization.

In this interview, Calabrio CEO Tom Goodmanson discusses reasons behind the rapid growth in speech analytics, how companies are using analytics on spoken content, and how the underlying data can be used by other departments, such as sales and marketing. "Once in place, speech analytics can help companies parse through customer conversations and apply the revealed data in a timely fashion for accelerated, data-driven decision making," Goodmanson says.

Upside: The speech analytics market is forecast to grow rapidly in the next few years -- from $589.2 million in 2015 to $1.60 billion by 2020, a compound annual growth rate of 22 percent. What are some of the factors contributing to this projected growth?

Tom Goodmanson: There are several factors that are changing how companies look at speech analytics, but the primary factor is technology.

Companies now realize there is a huge amount of rich data inside customer conversations, data that can help them determine customer sentiment, both positive and negative. Now that [analytics] capabilities have improved greatly, companies can understand nuances or meanings within conversations in ways that they couldn't before. Speech analytics can also reveal trends, which makes the technology useful to companies that are trying to connect with their customers on a deeper level.

Another important factor is that companies are using speech analytics to uncover customer intelligence for use outside the contact center. Analytics technology is being used by other departments, including sales and marketing, as part of an integrated strategy.

Companies are also putting more emphasis on customer relationships as part of a retention and growth strategy, and they're looking for any data to help. Speech analytics are giving them a window into the mind of the customer, particularly within contact center interactions, so they can adjust those strategies accordingly.

Are companies using speech analytics effectively already? If not, what are some of the challenges?

Some companies are using speech analytics effectively, but there is always room for improvement. Speech analytics requires constant monitoring and tweaking. In addition, many companies are currently listening more to the cognitive words in a conversation -- words that are most important to the sentence structure. However, companies must also focus on the value of so-called "filler words," which actually reflect customer sentiment.

For example, pronoun choice can indicate a lot. When a customer starts a conversation, he or she may say something like, "Yes, we've talked about this before." As the conversation progresses, he may say, "No, you haven't been able to fix this." The use of the word "we" is more inclusive, but the use of the word "you" may indicate that the customer is distancing himself from the brand. By understanding the use of those "throwaway words," brands can dig even deeper into customer sentiment.

Analyzing call center speech in particular seems like a great fit for speech analytics technology. How much is it being used?

It is a great fit, and there are contact centers using speech analytics very effectively both with phonetic and speech-to-text engines. This includes capturing conversations with customers and identifying key words or phrases in order to drill down on specific conversations. It's important because you're hearing what customers want -- right from their own mouths. Companies can't afford to miss out on this data.

Does speech analytics mean capturing an immense amount of data -- the spoken words -- before analyzing it? Is the sheer volume of data a constraint?

Companies are capturing thousands of conversations with customers every day. This equates to terabytes of data to analyze, but with the right analytics tools, it's not daunting. With today's technology, implementing speech analytics tools correctly can be half the battle. Once in place, speech analytics can help companies parse through customer conversations and apply the revealed data in a timely fashion for accelerated, data-driven decision making.

Can you give an example of how speech analytics for a call center would work? What might be analyzed, and what changes might be made based on that analysis?

In the contact center, historically, calls were initially recorded and put on the shelf, only to be accessed again based on a specific request. Now, companies can pinpoint keywords and correlate the customer calls with enterprise data. This equips companies to link customer engagement to business goals and share valuable insights across the organization.

For example, if a promotion is offered, companies can isolate and analyze the conversations surrounding that promotion. From there, they can determine if that promotion is working or not, and provide analysis to marketing.

This works well for customer insights; it also allows companies to monitor their contact center agents, to understand what they are doing during interactions. This gives companies insight into issues such as employee sentiment so they can improve employee retention.

What are some other areas and industries that are making use of speech analytics already?

Speech analytics can be used to determine if an issue with a customer is a marketing problem or an operational problem. With marketing problems, the company can determine whether customers understand the offer and adjust their approach accordingly. Sales can use speech analytics to identify cross- and up-sell opportunities and provide the right training to their teams to increase conversions.

Organizations can also find out which products or services customers are talking about most. That, in turn, informs the sales and marketing team about what customers really want. Speech analytics can also "dial in" on whether brand marketing efforts are resonating with customers. The research and development department can benefit from customer insights about product issues and feature requests.

With operational problems, speech analytics can be used to measure the extent of an issue. Rather than issue a blanket announcement of a problem to all customers, a company can use speech analytics to investigate and identify customers impacted by the problem.

For example, if a company discovers a billing issue, they can leverage customer interactions from the contact center to gauge the extent of the problem. If the billing issue only affected customers in a certain area, the company can communicate to only that subset of customers, minimizing the fallout from the problem.

Many industries can benefit from speech analytics technology. Heavily regulated industries such as banking, finance, insurance, and higher education have specific compliance and regulatory needs. They are at risk of fines if an employee says the wrong thing during an interaction. Speech analytics gives those organizations additional peace of mind by allowing them to flag instances of problem phrases and then coach employees to mitigate risk.

What are some of the challenges of using analytics on the kind of unstructured text found in speech?

The initial challenge is to get the voice of the customer into a format that can be read by machines. Although it's easy to capture the speech, it takes speech analytics tools to structure and unlock the insights contained within. The challenges include ensuring the precision of the results by tuning the tools so as not to deliver false positives. Fortunately, the technology is maturing and we are able to get better data all the time.

Can you describe some ways in which Calabrio is being used, in call centers or elsewhere?

Calabrio is being used in a variety of ways within the contact center, but it's also linking contact center data to the broader organization. Customers use phone, text, chat, or email to connect, and Calabrio is allowing organizations to roll all of that data up into one place to analyze it and improve the customer experience.

Companies can analyze the effectiveness of marketing campaigns, whether a product issue needs to be addressed, or any other indicator of customer sentiment. Customer calls are the one place where companies can hear the voice of the customer directly, and Calabrio gives them the listening tools to understand and adjust accordingly.

One of our retail customers is using speech analytics to quantify product complaints. The contact center is sharing that information with the merchandising department to prevent further customer frustration. Another customer, a university, analyzes 70,000 calls per day to identify compliance violations by their employees. They then use those findings to improve their training programs.

What's ahead for speech analytics?

Companies are always looking for a competitive edge, so we'll see expanded speech analytics adoption to better understand the customer. Companies are also seeking technology that will allow them to correlate the customer data from speech analytics with corporate data to identify opportunities to streamline operations. In addition, the technology is advancing to better leverage machine learning algorithms; the goal is to further develop cognitive analytics to get ahead of customer wants and needs.

TDWI Membership

Accelerate Your Projects,
and Your Career

TDWI Members have access to exclusive research reports, publications, communities and training.

Individual, Student, & Team memberships available.