Artificial Intelligence: Enabler and Accelerator for Digital Customer Experiences
Authors: Florian Stahl, Annika Hauser
Illustrating how companies can apply artificial intelligence in order to create an excellent customer experience
In this day and age, consumers no longer want a product or service to simply fulfill its basic functionality, like e.g., being able to get from A to B when purchasing a car (Homburg, Jozić, and Kuehnl 2017). In virtually every industry, product features are becoming easier to imitate by other brands, which makes it harder for companies to differentiate themselves through their products alone. Clearly, companies must seek a competitive edge elsewhere – and this is where superior customer experience comes into play (Bleier, Harmeling, and Palmatier 2019). Take Tesla, for example: Through AI, Tesla monitors everything that is going on in their cars. In doing so, Tesla is able to predict issues like part failures before the customer even knows that something is wrong. As a result, Tesla can immediately provide the necessary service to the customer upon occurrence of the actual failure, thereby ensuring a superior customer experience (Lobzhanidze).
The basics of Customer Experience
The concept of Customer Experience is ever-changing
Due to the ongoing digital transformation, we observe tremendous change regarding the concept of customer experience. When customer experience first became a relevant concept to the disciplines of marketing and sales, customers were viewed as passive entities who just responded to companies’ offerings. Today, in times of AI, the Internet of Things, and smart devices, customers actually interact with such devices, which brings whole new considerations to the concept of customer experience (Hoffman and Novak 2018). Digital advancements further enable companies to add experiences to virtually all touchpoints with their customers. Coca Cola, e.g., uses AI in China to bring interactivity to the purchase of drinks from vending machines: when customers buy a coke, they, for instance, receive a message on their smartphone containing a reminder to recycle the bottle. Recycled bottles then result in credits that can be shared with friends or spent on new purchases (Zaki 2019). (See also: How Chinese Companies deliver excellent Customer Experiences)
To get a better grasp on how different AI tools can deliver superior customer experiences, one should be familiar with the underlying customer needs; whether those needs are satisfied or not, determines if the customer evaluates an experience as positive or negative.
Four factors make up excellent Customer Experiences
Personalized Experiences
Because the ongoing digital transformation brings along countless possibilities for product and service design, customers increasingly desire experiences that match their personality (Bolton et al. 2018). This is especially relevant, as offerings from different companies become less differentiated: if customers cannot identify any other differences between products or services, they will choose the one that best matches their personality (Chung, Wedel, and Rust 2016).
Meaningful Interactions
Driven by technological advancements, customers seek experiences that satisfy a need for meaningful interactions. As mentioned before, in a world in which customers are connected to firms, products, or other customers through technology, they are no longer silent observers – instead, customers can and want to interact with other actors and actively participate in the value creation process (Bolton et al. 2018).
Immediacy
Another experiential customer need driven by technology is the need for immediacy. Today, it is easy for customers to access products and services – but also information about products and services – quickly through diverse channels. Due to this development, customer expectations have risen. This means that whenever customers have a question, they demand a prompt answer, or else they might just move on to another product or service (Parise, Guinan, and Kafka 2016).
Avoiding Decision Regret
Customers naturally seek to increase their overall well-being through avoiding decision regret. This can cause serious cognitive stress as customers seek to make the best possible choice regarding, for example, a product to purchase or a movie to watch. Customers are nowadays confronted with endless, often overwhelming information streams, which naturally increases the risk of fearing that a better consumption decision could have been made. Such decision regret will lower customers’ well-being and result in a negative customer experience (Kumar et al. 2019).
Bringing the factors together
Interplay of experiential customer needs (Source: Author’s own work)
The introduced factors together determine the quality of a customer's experience. This means that the strongest customer experience can be achieved if all four introduced experiential customer needs are fulfilled. Accordingly, the fulfillment of only one of the introduced needs doesn’t necessarily lead to experience improvements. Of course, the factors also influence each other: Where customers find a personalized experience, they likely have the possibility to avoid decision regret because the experience is personally tailored towards a customer's needs and therefore, in a personalized setting, often only those options are available that match the customer's behavior. It is helpful to keep this in mind when deciding which customer needs you intend to address.
AI tools which are relevant for creating superior customer experiences and how to achieve it?
To understand how AI can be leveraged for superior customer experiences, marketing and sales managers need to be able to distinguish between different AI tools. But before diving into specific applications, let’s first define what AI actually is: the term artificial intelligence refers to machines, systems, algorithms, or programs that are capable of performing intellectual tasks that would otherwise be performed by humans (Davenport et al. 2020). Machine learning, natural language processing, and neural networks are some key technologies that enable AI to perform those intellectual tasks (Davenport 2018) and collect and analyze data to learn from the insights. Based on these insights, AI can continuously and flexibly adapt (Davenport et al. 2020).
In a marketing context, AI allows companies to solve diverse marketing problems by making predictions based on historical customer data. Hereby, AI has proven to be faster and more accurate than humans in making predictions (Overgoor et al. 2019).
Recommender Systems
How Recommender Systems work
Companies deploy AI to personalize products or services, for instance through recommender systems which are for example used by Amazon or Netflix. Such recommender systems serve customers with recommendations that are based on their personal behavior. Here, the AI-application is based on machine learning algorithms that identify patterns from customer data (Davenport and Ronanki 2018). Recommender systems have the potential to meet the experiential needs of personalization, well-being (avoidance of decision regret), and – to some extent – interaction.
How Recommender Systems influence the Customer Experience
Personalized Experiences: It is quite clear how recommender systems fulfill a customer’s need for personalization: The machine learning algorithms powering recommender systems learn customers’ preferences and then adapt to them to recommend products that best suit each individual customer.
Avoiding Decision Regret: Recommender systems also address customers’ need for increased well-being through avoiding decision regret. When confronted with such a vast number of choices, customers fear regretting their decision later on, which can cause serious distress, or might even force customers to abandon the decision process. Here, recommender systems come in play and narrow down customers’ choices. This allows customers to process all the information and make a decision with greater confidence and less fear of decision regret (Kumar et al. 2019).
Meaningful Interactions: If recommendations are framed in a way that makes customers feel like the recommendation was made by another customer (e.g., by recommending products that were purchased by other consumers with similar preferences), recommender systems can give customers a sense of interaction. Although no “real“ interaction takes place, which is why this effect is limited, customers still subconsciously appreciate this semi-interaction with other customers and thereby feel good (Gai and Klesse 2019).
When a company uses AI to satisfy the customer needs of personalization, well-being through avoiding decision regret, and interaction, it can create a superior customer experience, because, on the one hand, customers feel like the company that makes appropriate recommendations knows, understands and cares about them and on the other hand, customers have a much more relaxed experience when it comes to selection processes of any kind (Dzyabura and Hauser 2019). (See also: How to implement and scale Customer Experience Excellence throu data-based Empathy)
Examples of Best-in-Class Recommender Systems
Many best-in-class examples of companies exist that successfully harness recommender systems to cater to their customers. Take Netflix or Amazon, for example: based on customers’ previous behavior, the firms successfully make recommendations to customers about what to watch or buy (Ansari, Li, and Zhang 2018).
Another example of a company that excites customers through an unparalleled use of recommender systems is Stitch Fix. Stitch Fix’s customers receive monthly deliveries containing pieces of clothing or accessories that match the customers’ fashion taste. Customers are not directly involved in the selection process. Instead, Stitch Fix determines the right product mix by applying AI to analyze customers’ social media data, data on past purchases, etc. to derive precise recommendations (WIRED).
Smart Agents
How Smart Agents work
AI can also be deployed in the form of (voice-controlled) smart agents. Such AI-driven smart devices can interact with customers or other smart devices through the internet and make autonomous decisions to adjust to context-related circumstances (Schweitzer et al. 2019). Like recommender systems, those agents rely on pattern recognition through machine learning algorithms. They further use natural language processing technologies to be able to interact with customers via voice or text (Davenport and Ronanki 2018).
How Smart Agents influence the Customer Experience
(Voice-controlled) smart agents are able to create superior customer experiences by fulfilling all four of the introduced experiential customer needs.
Personalized Experiences: Regarding personalization, smart agents exhibit an enormous potential: smart agents can adapt to customers’ behaviors and habits in a way that allows for a strong bond between the device and the customer.
Meaningful Interaction: (Voice-controlled) Smart agents are a very powerful tool when it comes to fulfilling customers’ need for interaction. Imagine, for instance, a customer’s verbal interaction with Amazon’s Alexa. During the interaction, the customer perceives Alexa to behave – to some extent – in a human manner. This makes the interaction enjoyable and allows the customer to add an interaction component to many daily routines like listening to music by asking Alexa to play a playlist (Verhoef et al. 2017).
Immediacy: Customers’ need for immediacy can also be effectively addressed by smart agents. As they are powered by AI, smart agents can react to customer requests in real-time and from any location or during any time of the day (Chouk and Mani 2019).
Avoiding Decision Regret: The need for increased well-being can be addressed by (voice-controlled) smart agents analogously to recommender systems. As smart agents adapt to customers’ preferences and behaviors, they can provide customers with tailored recommendations upon request, which will – due to the trusting relationship that often exists between customers and (voice-controlled) smart agents – lower the risk of decision regret (Kumar et al. 2019).
By fulfilling experiential customer needs as described, (voice-controlled) smart agents are not only able to create a strong bond with the customer that can result in a trusting relationship, but also make various everyday-tasks more enjoyable by adding meaningful, human-like interactions. Thus, (voice-controlled) smart agents help to boost a superior customer experience (Kumar et al. 2019).
Amazon as a Best-in-Class Example
One very successful voice-controlled smart agent is Amazon’s Alexa, which is able to perform different tasks upon the user’s verbal request (Hoffman and Novak 2018).
Amazon excels when it comes to connecting customers to its voice-controlled smart agent. Customers can use it to control their entire home, e.g. by using Alexa to control their lights, thermostats, etc. Hereby, Alexa uses its natural language processing capabilities to interact with a customer and then, through its AI-capabilities, learns from those interactions and adapts to customers’ needs. The customer experience gradually expands to more parts of customers' lives as the scope of Alexa's use becomes broader (Forbes).
Chatbots
How Chatbots work
In order to improve customers’ service experiences, many companies use AI in the form of chatbots. Chatbots imitate human conversations in order to interact with customers. Hereby, the conversation can take place via a textual chat or via a verbal phone call (Luo et al. 2019). Regarding the technology, chatbots function in a manner comparable to (voice-controlled) smart agents. However, the use of chatbots is usually limited to website visits and service calls (Davenport and Ronanki 2018).
How Chatbots influence the Customer Experience
Like (voice-controlled) smart agents, chatbots are able to fulfill all four experiential customer needs.
Peronalized Experiences: Regarding personalization, the AI that powers chatbots is able to fully adapt to customers’ needs when they interact with the chatbot during a website visit or service call (Komiak and Benbasat 2006).
Meaningful Interactions: Chatbots are also able to meet customers’ need for interaction, because they add a human-like interaction component to activities like online shopping, when providing the customer with support.
Immediacy: When it comes to satisfying customers’ need for immediacy, chatbots are particularly effective: the AI powering chatbots is constantly available, allowing to serve customers with immediate personalized answers or solutions (Tan, Wang, and Tan 2019).
Avoiding Decision Regret: Chatbots can help avoid decision regret in the same way as (voice-controlled) smart agents can, namely, by providing customers with appropriate product or service recommendations that are based on a customer’s preferences upon request (Kumar et al. 2019).
Concludingly, chatbots improve the customer experience by providing personalized, immediate support and by adding an experience component to activities that would otherwise not include one. Like this, they help customers navigate through (potentially overwhelming) situations in an enjoyable manner (Luo et al. 2019).
The Starbucks Barista delivers excellent Customer Experiences
Starbucks provides customers with assistance through the “Starbucks Barista”, an AI-powered chatbot that uses natural language processing to interact with customers. With each purchase a customer makes, the chatbot collects data and learns from it to adapt to the customer’s behavior. Already on their way to the shop, customers can tell the “Starbucks Barista” that they want to get a coffee. The chatbot then asks follow-up questions, makes personalized recommendations, etc. Customers have a fun experience, good service, and can pick up their beverage right away when they reach the Starbucks shop (Geekwire).
Analyzing Textual and Contextual Data
How to analyze Textual and Contectual Data
Especially for marketing purposes, AI can be used in the form of supervised machine learning techniques for the analysis of textual and other contextual data that are able to detect response-worthy electronic word-of-mouth on social media. Such machine learning algorithms are trained on data that is pre-classified as response-worthy or not. They then learn to predict whether large amounts of unclassified data are response-worthy. These techniques capture the context in which a social media post or comment was written and how relevant the content is to the business. This enables companies to identify relevant comments or posts and, in turn, create meaningful interactions with customers (Vermeer et al. 2019). (See also: An invitation to dance: The optimal Customer Experience in Customer Service)
How the analysis of textual and contextual Data influences Customer Experience
The described machine learning techniques can satisfy customers’ needs for interaction and immediacy. Driven by the current social media hype, many customers feel the need to share their opinions with companies through social media channels – via private chats, but also publicly visible through comments or posts (Yadav and Pavlou 2020).
Meaningful Interaction: With respect to the need for interaction, the introduced machine learning techniques can help companies identify which posts they should respond to in order to create meaningful interactions with their customers.
Immediacy: Similarly, also the customers’ need for immediacy canbe addressed, because the AI-powered detection of response-worthy social media contributions allows companies to provide an answer shortly after the contribution has been made (Vermeer et al. 2019).
As a result, customers feel appreciated, because the respective company hears their “voice” and provides an immediate, public answer. That leads to a much better customer experience.
How Wendy’s leveraged the Benefits of Textual and Contextual Data Analysis
In 2017, fast food chain Wendy's impressively demonstrated the power that a response to a seemingly insignificant tweet (Twitter post) can have: A customer addressed Wendy’s in a tweet asking how many retweets he would need to get free nuggets. Without any form of textual and contextual analysis, such a tweet would just drown in the flood of tweets companies receive each day. Nevertheless, Wendy’s actually tweeted back and told the customer that he would get free nuggets if he achieved 18 million retweets. Even if the customer didn’t reach 18 million retweets, his tweet still became to most retweeted tweet ever. Not only did Wendy’s get a lot of publicity, but the customer and many other customers following the challenge got a great customer experience (Oberlo).
Conclusion
Comparing the different tools and methods
Comparison of the different AI tools and their impact on Customer Experience
From the previous presentation of different AI applications and their ability to fulfill experiential customer needs, it is clear that (voice-controlled) smart agents and chatbots are the most powerful tools to boost customer experiences with respect to all considered experiential needs.
It should hereby be noted that (voice-controlled) smart agents offer the greater capability of creating a strong bond with customers, but chatbot solutions, on the other hand, can already be deployed at a much lower level of complexity.
In the e-commerce context, recommender systems are particularly powerful, because they give companies the ability to turn activities such as online shopping from a headache into a fun experience. The recommender system’s design and the quality of its recommendations are hereby crucial in order to gain customers’ trust.
Machine learning techniques to detect response-worthy electronic word-of-mouth might seem less powerful in creating a superior customer experience compared to the other introduced AI tools. However, managers need to keep in mind that these techniques are comparatively easy to implement and yet offer huge added value when it comes to enabling an enjoyable communication between a firm and its customers.
Some words on Implementation
In general, the development of AI applications requires a lot of expertise. Irrespective of industry or company size, managers who want to create superior customer experiences should thus consider
1) what customer needs they want to satisfy during the customer journey, and
2) what capabilities are available to develop or buy AI applications that fulfill those needs.
Therefore, despite the option of purchasing complete solutions from AI-specialized firms, companies should be prepared to face substantial difficulties when implementing AI applications. (See also: The Transformation to a CDXE Culture)