Augmented Intelligence
Dec 13, 2017

AI sounds the death knell for traditional customer satisfaction surveys

The exciting new realm of artificial intelligence (AI) is having a real impact on the world, with more and more people talking about the promise it offers—and the dangers it invites. True, AI powers driverless cars, improves medical technology, creates better customer experiences and turbocharges business. But debates also rage about how AI could create mayhem on earth: We've all seen sci-fi movies or read novels that depict it as a disaster. At any rate, it's safe to say that AI has become a fact of life and is getting more relevant by the day. We see commercial deployments and large-scale AI adoption across industries catching on like wildfire.

AI's scope and evolution have been particularly formidable in customer experience, an area where it is actively being used to gain competitive advantage. Several high-profile experiments highlight how companies are applying it to better understand customer behavior and improve customer experience. 

The problem with surveys

In the past few years a range of customer relationship management (CRM) tools have been helping companies measure customer satisfaction through surveys such as net promoter score (NPS) and customer satisfaction score (CSAT). No question, these initiatives foster customer retention, deliver channel feedback, and reveal marketing opportunities. But in the past, paper-based surveys typically had response rates of 8%-10%. For e-surveys, the rate was 5%-15%. Today, those rates have fallen to an all-time low of less than 2%. What's more, CSAT scores, for one, sometimes don't accurately reflect user sentiment because they rely on point-in-time ratings and don't convey the broader picture.

And no matter how widely or exhaustively companies circulate their surveys, they never seem to get a holistic sense of customer experience. Let's face it. Customers don't like questionnaires and will avoid them unless someone nudges them to fill them out. Questionnaires can even do harm. Suppose, for instance, a customer calls a bank to complain about an incorrect debit. An empathetic customer support executive promises to correct the error and credit the customer's account—but somehow, that doesn't happen. Yet the customer gets a barrage of messages and emails asking them whether the bank resolved the issue and how they'd rate their interaction with the bank's representative. You can imagine the customer's exasperation. What seemed like a risk-free survey has now caused significant reputational damage. 

A more sophisticated approach

 The next step in measuring customer satisfaction is more sophisticated. It calls for the latest in AI technology, already on stream and gaining prominence. Differentiating factors such as speech recognition analytics, loyalty analytics, social media analytics, and a suite of other tools are upping the game and helping businesses keep a more granular check on customer experience. 

Scientifically based customer effort scores are the key

The lesson here is that no single parameter provides a cumulative view of customer experience at different touchpoints. Only when KPIs link to other KPIs can the insights they deliver be exponential. They can show, for example, that the effort a customer expends to resolve an issue is just as important as the satisfaction score. If the customer had to go through numerous channels and couldn't settle a problem in three steps, that's bad for your reputation. Measuring that dissatisfaction is the concept behind the customer effort score (CES), another tool in the customer relationship management arsenal.

The Harvard Business Review first posited the idea of CES—that is, the number of steps a customer must take to solve a problem—in the article Stop Trying to Delight Your Customers. HBR found that a CES was a more accurate predictor of customer loyalty than a CSAT. The article recommended that companies introduce the effort score into customer feedback programs—something that would radically change the survey process.

Key questions about customer effort

Two important questions you should ask about CES: How can you use AI to capture and measure feedback for actionable insights? And will these measures provide a holistic view of the work it took for customers to get their issues resolved? At Genpact, we advise our clients to integrate scientific and accurate measurement into CES calculation with a feedback system. The reason:

  • You can capture large sample sizes using platforms that take advantage of big data to get granular details and derive clear insights about how long it takes customers to resolve issues 
  • You can segment areas where customer effort is high, then eliminate bottlenecks and introduce effort-reduction initiatives 
  • You get a view of all customer touchpoints, so your agents and team can respond promptly

Scientific CES delivers real value by improving the overall efficiency of support operations. It's proactive. It identifies end-user pain points for quick action. And Genpact Cora—an AI-based open-source platform—can accelerate this digital transformation. It's a modular, interconnected mesh of flexible digital technologies that helps large global companies reframe and solve their most pressing business issues.

Genpact Cora can specifically help with CES by suggesting priorities based on feedback from a user effort index. This well-defined platform also helps to identify various issues, gaps, or channels causing high levels of effort on the part of users. What's more, AI connects information from independent channels and collates their scores to act as an ongoing insight-generation engine, so management can make better decisions. Finally, the underlying model has multiple analytical layers empowered by AI techniques.

Figure 1: View of all customer touchpoints

Enhancing customer service

We've studied customer effort across many channels and in a range of businesses, so we know how to help our customers make strategic decisions. Some outcomes you can expect include:

  • Channel deflection strategies that let you eliminate ones that don't work or that your customers don't use and replace them with more efficient ones
  • A proactive way to address specific issues, time of call, or geography based on platform predictions
  • Reduction in customer and agent efforts 
  • Improvement in agent feedback mechanism 

At Genpact, we use AI to provide a single view of all key metrics—and we do it in an iterative way, with algorithms based on effort and learning. AI and machine learning let you live through the customer experience across all touchpoints. At the same time, they embed analytics in all your core processes to create scalability and data-driven decision-making. 

The bottom line: AI can take the pain out of customer surveys

To get the most out of measuring customer effort, you must clearly understand the key drivers of customer dissatisfaction and churn. AI provides the right information—and a solid picture of a customer's level of loyalty or disenchantment. It can also help map brands to new customer engagement methods. This increases conversions and provides an ongoing measure of customer satisfaction against key business parameters. AI engines are smart enough to learn from customer behavior. In fact, no other approach has the science behind it to build great customer experience and loyalty the way AI does.

The blog is authored by Rohit Tandon, chief analytics officer, Genpact with inputs from Sudha Bhat, assistant vice president, customer analytics, Genpact.

About the author

Rohit Tandon

Rohit Tandon

Business Leader, Analytics

Follow Rohit Tandon on LinkedIn