The most common methods of tracking customer sentiments has a big blind spot: They can’t pick up on important emotional responses. As a result, qualitative surveys, like Net Promoter Score, end up missing critically important feedback. Even if they provide a positive score, customers often reveal their true thoughts and feelings in the open-ended comment boxes typically provided at the end of surveys, and AI can help companies make use of this valuable data to better predict customer behavior.
Specifically, there are six benefits for adopting AI to analyze this feedback: It can
1) Show you what you’re missing in your qualitative surveys,
2) Help train your employees based on what’s actually important to customers,
3) Determine root causes of problems,
4) Capture customers’ responses in real time,
5) Spot and prevent declines in sales, and
6) Prioritize actions to improve customer experience.
In order to succeed, firms need to understand what their customers are thinking and feeling. Companies spend huge amounts of time and money in efforts to get to know their customers better. But despite this hefty investment, most firms are not very good at listening to customers. It’s not for lack of trying, though – the tools they’re using and what they’re trying to measure may just not be up to the task. Our research shows that the two most widely used measures, customer satisfaction (CSAT) and Net Promoter Scores (NPS), fail to tell companies what customers really think and feel, and can even mask serious problems.
For years, quantitative surveys have been the industry standard. They ask customers a single question: On a scale of 0-10, how likely are you satisfied with this company’s product or service? Or how likely are you to recommend this product to a friend or colleague? While these surveys are resource intensive, and customers are finding them increasingly intrusive and are becoming less inclined to participate, they’ve remained a core piece of companies’ strategy for understanding their customers.
The problem is these surveys can’t pick up important emotional responses and end up missing critically important feedback as a result. In our research, we found that customers often score firms highly in surveys even when they experience significant problems with their products or services – a vitally important response that they miss. And by masking significant customer dissatisfaction, these surveys can cause firms to lose customers without knowing why.
There is, however, a goldmine of good data if you know where to look and how to analyze it. Customers often reveal their true thoughts and feelings in the open-ended comment boxes typically provided at the end of surveys. In general, the content of these comments offers a much more reliable predictor of a customer’s behavior. Yet, these are often ignored, and if used at all, are typically used after the scores are computed.
The good news is that most companies have the power to correct this oversight relatively quickly. We developed an AI-driven approach that practitioners can use as a model to adjust their customer feedback processes accordingly.
HOW AI CAN HELP
It’s easy to see why quantitative surveys became popular: They’re a way to ask a huge number of customers how they felt. Qualitative approaches, like focus groups or manually reading and analyzing customer feedback, were too labor intensive to scale. Now, technology has changed what’s possible, and tactics need to catch up.
The first and most significant change firms should make is to flip where they’re investing in their analysis of customer sentiments. They should start with the qualitative comments, and then turn to the results of their quantitative surveys. If they have the right tools to analyze the qualitative data (e.g., customer relationship management systems, social media, customer reviews, emails, call center notes, chatbots, etc.), firms could even consider ditching quantitative surveys altogether, as these make it possible to hear what customers are thinking and feeling across multiple touchpoints in real-time.
This is where AI models and tools can help. AI tools are not yet widely adopted by marketers and customer experience managers, and those that are available tend to indicate only positive or negative sentiments. In our research, we used a customer-focused framework to extract and map keywords representing the customer experience (CX) to the following dimensions: resources (e.g., knowledge, system, product, skills, etc.); activities (e.g., fixing, ordering, service delivery, etc.); context, or situations affect the experience (e.g., weekend); interactions (e.g. calling, chatting, etc.); and customer role (e.g., provides suggestions or neutral). We then identify both customer emotions (joy, love, sadness, anger, and surprise) and cognitive responses (compliments, complaints, and suggestions) at touchpoints.
For example, one customer gave 10 out of 10 in CSAT score. However, they also offered the following comment: “The only thing that we were a bit disappointed with is to do with repairs. It seems that every time they come out it’s over $1,000 in service. The fitters seem to be struggling with diagnosing the issue and it always seems to be more expensive.” We applied linguistics-based natural language processing (NLP) approach to extract and map keywords in this comment. For example, “repairs” is associated to “touchpoints,” “fitters” is mapped to resources, “diagnosing the issue” is classified under activities, “a bit disappointed” is considered a sadness emotion, and terms such as “over $1,000,” “struggling,” “more expensive,” are categorized under complaints.
Finally, our AI generates and converts key features into predictive variables that can train the model to predict whether customers are satisfied, neutral, or have a complaint, without using quantitative survey scores.
AI algorithms can capture specialized vocabulary used by customers and combine their views expressed in their own words with traditional rating scales to obtain deep insights. These insights can directly shape both short-term and long-term actions to retain customers.
We tested our AI tool on longitudinal customer experience data collected by four multinational service providers – a data set of roughly 30,000 comments. These firms ask customers to rate their services using the traditional CSAT and NPS metrics and tack on an open-ended final question. While the data we used came specifically from these survey questions, data from any kind of qualitative source would work. Here’s what we learned.
SIX KEY BENEFITS FROM USING AI
In our research, we found that AI can transform how firms think about and measure customer experience, but six benefits in particular stood out.
AI Can Show You What You’re Missing
Companies often misjudge what their customers really want. We found that the touchpoints that customers really care about may not be the ones that firms expect. Importantly, this AI-driven qualitative approach can show you what you’re missing, and therefore how to fix it.
For example, one firm was focusing only on sales, parts, field service, and workshop touchpoints, but customers generally considered credit financing and invoicing touchpoints to be more critical in their interaction with the firm. As a result of this insight, the firm could redirect its resources.
Train Your Employees Based On What’s Actually Important To Customers
Understanding how your customers work with your firm allows you to build a customized training program to educate employees on how to empathize more with customers, care about their issues, and to interact with them seamlessly.
For example, our model highlighted that one firm’s employees were often inflexible and showed little care when faced with customers’ complaints. Based on this insight, the firm trained employees in customer experience workshops to deliver key messages about customer care, customer empathy, service recovery strategies (what to do when things go wrong), and taking corrective actions. By following these customer experience actions, firms saw an increase in customer satisfaction, and an improvement in retention.
Determine Root Causes
To fix a problem, you need to understand it. When it comes to customer experience, companies can use AI-produced insights to glean not only where there are problems, but also what’s causing them.
In one case, communication was a major pain point. Insights gained were used to repair relationships with customers who were identified as likely to defect. The company undertook decisive actions. First account managers started to follow up with these identified customers to really get to know what their concerns were. Then the firm invited key customers to a corporate event to discuss in one-on-one meetings the reasons for the service failures.
Capture Customers’ Emotional And Cognitive Responses In Real Time
Firms should capture how customers feel about the service through discrete emotions – joy, love, surprise, anger, sadness, and fear – and extract cognitive responses, conceptualized through customer evaluations (e.g., complaints, compliments, and suggestions) in real time. It is important to capture real-time feedback as emotional and cognitive responses can dissipate over time and details of the interaction are likely to be forgotten. AI analysis allows firms rethink their current customer experience measurement program.
For instance, one of the firms we worked with is piloting three critical touchpoints and embedded feedback mechanisms into each of them to analyze data in real time using our AI model.
Spot And Prevent Decreasing Sales
Firms can segment customers based on their monetary value by using NPS with customers’ emotional responses spotting decreasing sales. In one of the case companies, we identified customers who, although giving high CSAT or NPS scores, were at risk of defecting due to historical issues. We demonstrated to the firm that if these so-called “satisfied” customers defected they were likely to cost them around $6 million in lost sales. This insight could alert firms to any decline in sales, and help them reduce costs associated with losing customers and acquiring new customers. Spotting when a customer has slipped to a lower category score, enabled the firm to interfere to avoid losing that customer.
Prioritize Actions To Improve Customer Experience
Finally, firms can use these insights to diagnose the underlying factors causing pain for customers and then prioritize which root causes need attention. This enables managers to stop doing certain actions (complaints), start doing new actions (suggestions), and continue doing actions (compliments).
This process can be codified and automated, so companies can see in real-time how particular areas are performing, drill down, and intervene on any emerging issues. The analysis also gives employees a view across the entire journey, enabling employees across the organization to have the same view of the customer, so that if problems arise all frontline employees are able to see what has happened and act accordingly.
Customer experience is now the major differentiation between competitors. As many customers today use smart, real-time services and friendly apps, firms can increasingly gather more real-time verbatim data about customers’ journeys instead of relying on simplified, single-metric ways of measuring customer experience. Combining these insights from direct customer comments, with analysis of customer transactions, and other sources can provide companies with a bespoke 360-degree view of the customer experience.
By implementing an AI-driven model like the one we’ve described above, firms can monitor the customer experience in real time and generate insights which would allow service providers to provide a seamless customer experience and intervene in a timely manner for effective service recovery. Hence, organizations can use data stemming not only from their own touchpoints but also from external touchpoints in the digital, physical, and social channels with the primary goals of continuously and proactively adopting customer experience to retain customers and achieve customer loyalty and long-term growth.
originally posted on hbr.org by Mohamed Zaki, Janet R. McColl-Kennedy, and Andy Neely
About Authors:
Mohamed Zaki is Deputy Director of the Cambridge Service Alliance (CSA) at the University of Cambridge. His research interests lie in developing novel machine learning methods to manage and measure customer experience and predict customer loyalty. He has published widely in highly rated academic journals has received international awards (2016 and 2017) from the Marketing Science Institute on customer experience analytics and digital customer initiatives.
Janet R. McColl-Kennedy is Professor of Marketing at the UQ Business School, Director of Research at the UQ Business School, The University of Queensland, Brisbane, Australia. She is a Visiting Professor at the University of Cambridge. Her research interests focus on customer experience management and includes service recovery, customer complaining behavior, customer rage and customer value co-creation.
Andy Neely founded the Cambridge Service Alliance (CSA), a leading centre of research into complex service delivery, in 2010. At the CSA, Cambridge academics work with global companies to solve challenging business issues. Professor Neely is currently Pro-Vice-Chancellor for Business and Enterprise and Director of the Centre for Digital Built Britain at the University of Cambridge. In 2018 he was elected a Fellow of the Royal Academy of Engineering in 2018.
