How to take QA call auditing to the next level
Once you automate your QA with conversation analytics and capture information from 100% of your customer interactions, shift your QA team’s focus to higher
Sara Yonker
February 6, 2023
Caitlin Jordan
October 27, 2020
CX leaders all recognize the importance of a robust structured voice of customer (VoC) data collection program. It’s the bedrock of any good customer listening approach.
But progressive companies are finding that adding unstructured data to their listening programs can help them surface insights faster, delve more deeply into friction points along the customer journey, and surface innovation opportunities that are sometimes missed when strictly relying on surveys.
Unstructured data—or what’s sometimes called “ambient” or “found” data—is data that is collected during the normal course of business. Think recorded phone conversations, chat interactions or case management data. It can also be data left by customers voluntarily, such as social media comments, user reviews, etc.
Unstructured data is quickly becoming an increasingly important part of a successful listening program.
Matt Dixon, Tethr’s Chief Product & Research Officer, recently sat down with Luke Williams, SVP & Senior Principal XM Catalyst at Qualtrics XM Institute, to discuss why that is, and how CX leaders can implement it into their own programs. To watch the full discussion, click here.
What follows below is an excerpt from their conversation.
Matt: This conversational analysis is showing to be pretty powerful because it really elevates the call center leader into more of a “listening post” for the enterprise.
First, we’re working with a number of companies to drive closed-loop actions after bad customer interactions. We can detect which calls went well and which went poorly and then Qualtrics customers can action this voice data taking the insights gleaned from voice conversations and automatically bring them back into other parts of the organization.
Second, customers are looking for ways to modernize their Quality Assurance programs in the call center. We’re able to help agents see exactly what behaviors, actions and language they need to be using (or avoiding) to create easy, low-effort customer interactions—which we know from our research leads to customers becoming more inclined to stay, spend more and advocate for the brand.
Finally, companies are using unstructured conversational data to spot pervasive, widespread “effort drivers” in their business—things that don’t really have to do with the call center agent but are problems that exist “upstream” in the business such as poor digital experiences, product defects or unclear differentiation vis-a-vis competitors.
Matt: There’s three clear benefits.
First, there are immediate productivity gains that come from reducing QA expenses, eliminating issues that drive escalations and repeat contacts and working with the business to solve problems upstream so that they never wind up in the call center.
Second, we’re seeing a lot of quality gains. Customers are tracking steady improvements in agent usage of low-effort skills and behaviors which we’re seeing reflected in better post-call survey scores and, ultimately, greater customer satisfaction.
Finally, customers are seeing much better sales conversion results since we can help them to really spot the things that the star sellers are doing that can be replicated across the entire call center.
Matt: Tonal sentiment analysis—looking at voice inflection and other audio characteristics—isn’t a terribly accurate way to address any of the use cases we’re talking about here.
Let me give you an example: a company may be wondering why they saw an uptick in call volume on a particular day and, if they’re using tonal sentiment analysis, they can identify that there was a spike in customers who expressed anger or frustration when they called.
But they can’t make a business decision on that. They still need to figure out why customers were so upset. The only way to really surface actionable business insight is to study the syntax of the conversation—literally, what is being said.
Anybody can transcribe their calls and turn them into a mountain of text and then just search for keywords—whether that’s looking for examples of foul language, product names, competitor mentions, etc. But that’s a pretty crude approach and it doesn’t capture much nuance in the conversation.
For instance, it’s not going to tell you whether a customer is saying they plan to switch to a competitor, whether they’re asking how your product compares to a competitor’s or whether they’re saying they used to be a customer of your competitor but they’re so glad they switched to you.
The true meaning of conversational analysis—and the actionable business insight—comes when you can understand the full context and nuance of what’s being said. For example, if there's a bad conversation, knowing this so that action can be taken is vital. Analysis will look at where in the conversation it turned bad, how many times this has happened, how the agent responded, etc..
Then it produces a single score on a scale of 0 - 10. Zero being a really, bad, difficult and high-effort conversation. Ten being an excellent, easy, low-effort conversation.
But upon knowing that information, successful programs automatically alert stakeholders that an issue exists, why it exists, enable them to take action to rectify it. Which is why the partnership with Tethr and Qualtrics is so valuable, because it combines the two.
Want more? You can view the entire conversation between Matt and Luke here.
Interested in learning more about VoiceIQ and how Qualtrics and Tethr can help you surface and act on insights from customer conversations? Request a demo here to learn more.