Customer Experience: Today’s Most Crucial KPI

Over the past decade, customer experience has risen to the top of CEO and board agendas. In fact, one of the world’s most ubiquitous corporations, Amazon, counts “customer obsession” as its guiding principle. Seeking to mimic the giant’s billion-dollar success, companies are pivoting to prioritizing excellent customer experience. According to McKinsey & Company’s “The State of Customer Care in 2022,” customer experience improvement is the fastest-growing priority among customer care leaders. 

In years past, the modus operandi of customer success teams was to decrease call volume in contact centers, lower customer service costs, and deflect customer contact. Now, companies are realizing that each touchpoint with the customer is an amazing opportunity to boost customer lifetime value and drive word-of-mouth recommendations. Heading off negative experiences with attentive customer service could save companies major hits to their brand reputation.

If leadership decides to establish customer experience as a key performance indicator, the next question becomes: “How do you measure it?” Many companies can measure long-term metrics, such as customer churn, customer lifetime value, and sales conversion. What is of higher urgency is the identification of leading indicators that teams can use to prevent churn and increase sales conversions.

But how can teams obtain such metrics?  

The Perils of the Customer Satisfaction Survey 

The most obvious answer is: Just ask your customers. However, the obvious answer may not be the best one. The manifestation of “just ask your customer” is the ever-popular customer satisfaction survey. Unfortunately, survey response rates are low, typically in the range of 10 to 30 percent. The few responses you do gather are likely biased. The customers you hear from tend to be the ones who have a reason to give feedback (a very negative or a very positive experience), the ones who have time to give feedback, or those who think they may earn a prize in exchange for a positive submission.

Moreover, surveys are challenging in that they rarely provide actionable feedback. They are often conducted days after a customer service interaction. Plus, agents don’t see the survey results. Though, given the above challenges, that’s not a loss.

The unreliability of satisfaction surveys means it’s difficult for contact centers to use the flawed data they collect to drive performance improvements. Instead, call centers must rely on other measures of success, such as average handle time.

Using Tone to Understand Customer Satisfaction 

Until recent AI and call center technology advancements (more on that later), capturing the tone of a customer interaction wasn’t possible. But this is now changing rapidly.

Most of us have encountered the argument that body language is more important in communication than the words said. This stems from Mehrabian’s 7-38-55 communication model, which proposes body language accounts for 55 percent of the communication, tone for 38 percent, and the spoken word for 7 percent.

The latest research confirms that Mehrabian was correct in asserting the importance of tone but underrated its magnitude. Michael Kraus of Yale University recently found that audio may be a more powerful tool for estimating sentiment than even video.

Whether it is truly body language or tone that has the most predictive power is ultimately not a pressing concern in contact centers, where video calls account for a tiny fraction of the calls. The critical insight is that tone provides a very interesting avenue for understanding customer satisfaction, much more so than the text-based solutions on the market today. 

AI to the Rescue? 

With artificial intelligence solutions hitting the mainstream, it’s no surprise that contact centers and technology companies are hurdling the lowered adoption bar and integrating AI-powered analytics solutions. Many call centers have employed speech-to-text technologies that convert audio conversations to text to also better understand customer sentiment. Yet these attempts have had somewhat limited success due to the spoken word only accounting for 7 percent of the communication content as described above.

Recent breakthroughs based on the same model architectures that have given the world amazing tools like ChatGPT and DALL-E are making waves in the customer service realm. For instance, leaders can gather what customers are calling about, what issues they are having and whether the agents are using the expected language and following scripts. Additionally, Björn Schuller, the top-cited professor in voice AI, and his team at audEERING has delivered the first tone-based models that are truly scalable and can detect sentiment accurately and robustly without biases. State-of-the-art AI models can understand customer sentiment based on tone, not the words uttered. The accuracy of some AI can match—and in some cases surpass—the skills of a trained psychologist listening to every call and rating whether the customer was satisfied in the end.

Evaluating customer satisfaction will probably always remain in the crosshairs between science and art, but the arrow is moving quickly toward science. Given that this is the most important metric in contact centers, company leaders should continue to push progress and strive for better measurements. With the advent of tone-based models, we are truly making strides and have—at the very least—a complementary tool, but possibly also a replacement for survey or text-based tools.

Anders Hvelplund is senior vice-president, Call-Centric BU & Global Services, at Jabra.