Customer relationships evolve over time; they may be influenced by external factors — such as technological developments — and may take new forms. In the past decade, for instance, consumers converse and interact much more frequently and extensively through social media networks, and thereby their relationships with companies and brands find greater expression in those virtual-online platforms and channels (i.e., adding blogs, micro-blogs, and chatting). In recent years we have seen a rise in the capabilities of artificial intelligence (AI) and their applications for different purposes (e.g., behavioural predictions, recommendations, interactions). In particular, applications of enhanced AI capabilities can, and already do, affect several aspects of customer relationships.
Barak Libai and his (six) colleagues (2020 ) delve into the emerging domain of AI-CRM — implementing AI in the area of customer relationship management (CRM). The researchers describe how AI technologies (e.g., analytic, operational) can influence the performance of functions or tasks in CRM and discuss their practical implications and broader consequences (e.g., personal, social, legal-regulatory, and ethical). They distinguish two critical capabilities enabled by AI: (a) leveraging big customer data, and (b) communicating, understanding, and creating the way humans do. Those capabilities are utitilised in three key tasks in the context of managing customer relationships: (1) customer acquisition; (2) customer development; and (3) customer retention. In the view of Libai et al. of AI-CRM, when combining the notions of AI and CRM, “any CRM system exhibiting sufficiently flexible adaptation can be labeled an artificially intelligent CRM system or AI-CRM” (p. 45, boldface added).
Through automated and advanced analyses of large volumes of customer data, and with high variety of data, companies can reach, for example, better abilities to personalise product offerings to customers, customise service procedures, and adjust interactions with customers. But leveraging real-time and rich data on differences between customers (e.g., in preferences and price sensitivity) can lead to increased discrimination and greater social inequality among consumers. The emergence of AI is likely to change the nature or form of relationships and relationship marketing: it may create new possibilities but also raise or emphasise problematic issues; AI may render relationship marketing more accurate and scalable, yet also more discriminating.
Libai et al. address in particular the criterion of customer lifetime value (CLV): they argue that companies would be more likely with the aid of AI to provide more favourable treatment to customers with higher CLV as opposed to treatment given to lower CLV customers (e.g., higher-grade versus lower-grade product or service, more versus less appealing price offers and deals). The power of data and methods of AI (e.g., machine learning: simple and deep artificial neural networks) can give companies a stronger ability to do so. Customers who are either less financially capable to make large expenses or less willing to be committed to a given company may find themselves in a greater disadvantage in their relationship with the company compared with higher value customers, and they may not be even aware that they are being disadvantaged. Consumers who rely on the aid of AI-enabled services and devices could face another consequence: feeling that they lose some of their autonomy (as in making decisions), which in case of being less satisfied with the outcome this may increase the sense of consumers that they are being manipulated or discriminated against.
A crucial benefit to be gained from leveraging big customer data is the knowledge companies can acquire from learning: about customers’ preferences, actions, attitudes, and even emotional expressions. The data can arrive from a myriad of sources: tracking online behaviour, structured survey questionnaires, social media, cameras and sensors, and more. AI methods and models can be trained to analyse and learn from both structured and unstructured data (e.g., free verbatim, sounds, images). Libai and colleagues suggest that of the three Big Data Vs (volume, variety, and velocity), variety is “the strongest driver of competitive advantage”. Variety is measured by the breadth and scope of the customer database, and the more types of data are included, “the more opportunities there are for discovering associations therein” (p. 46). They note, however, that companies need to combine and integrate data from internal (company-based) and external (third-party entities) origins in order to obtain a more complete picture of customers and the market (e.g., existing & prospect customers, individual-level behaviour and segment-level attributes); companies further need to appreciate the value of data exchange through partnerships with other entities. This consideration may prove especially vital and beneficial in the context of customer acquisition (e.g., selecting successfully prospect customers predicted to deliver higher CLV).
In the context of customer retention (extended duration of lifetime) and development (greater value-profit derived per period), Libai et al. highlight a shift in focus from customer loyalty to habits. Customers’ habits are acknowledged as more characteristic of consumer behaviour, a driver of market success, and which AI can more easily draw on and leverage. (Note: this approach no longer requires the establishment of attitudinal loyalty, just the prevalence of strong habits.) Artificial intelligence can help to fulfill two functions in this regard: (1) ‘teach’ and form new habits for customers to practise, and (2) reinforce, reward and enhance existing habits.
Habit forming, as the authors explain, is governed by automaticity and requires minimal cognitive attention for performing frequently repeated behaviour. For example, sensors included in Internet-of-Things (IoT) equipment and devices can collect and analyse data and guide the user on actions to be taken (e.g., refilling products in the refrigerator, cooking in the oven, adjusting climate conditions in the house). Additionally, virtual assistants like Amazon’s Alexa (mediated via Echo) can give easy-to-follow recommendations to the user on products to buy (e.g., clothing, food items for a recipe) or services to order (e.g., flight tickets and hotel room reservations when planning a vacation) — the more a consumer-user is satisfied with the outcomes and happy with the less effort needed on his or her part to make purchase decisions, he or she will be more likely to adopt this way of purchasing as a regular habit. Making habits of current customers better rooted means higher switching costs, and hence Libai and his colleagues expect that companies will have the advantage through utilisation of AI-CRM to increase switching costs and ‘protect’ their customers from churning than for competitors to offer ways to decrease switching costs and attract new customers to them.
Also, in the course of customer development and retention, personalisation may prioritise higher CLV — that is, customers with higher lifetime value can expect to receive better-fitting product offerings, more attractive buying opportunities, and improved personalised services, compared with customers exhibiting lower lifetime value. In other words, AI-CRM can help firms to selectively develop and retain customers, such as through personalisation that works more to benefit customers with high CLV. Of three forms of artificial intelligence identified by Huang and Rust — namely Mechanical AI, Thinking AI, and Feeling AI — the emotion-oriented form of feeling AI is assigned by them as having the greater role to play in the personalisation of relationships . Huang and Rust suggest that capabilities of feeling AI are best suited to provide ‘relationalization’ benefits: While thinking AI capabilities are engaged in constructing personalisation features and benefits (utilitarian, e.g., through recommendations, smoothing shopping journeys), feeling AI personalises relationships (emotional, connecting with the customers). The relational benefits may be delivered through interaction, communication, and by leveraging moods and emotions in marketing functions. Furthermore, the applications of feeling AI seek the goal of relational benefits when CLV is high.
Consumers may learn and develop their own ways to navigate in the face of AI interventions and applications. The capabilities of AI-CRM may enable firms to identify customers and detect differences or changes in their behaviour and preferences quickly and in real-time, and to provide rapid personalised responses. However, experienced consumers may also be able to detect some of the rules used by companies employing AI-CRM, and to make them work to their benefit. Customers who feel disadvantaged may provide false information to mislead and ‘penalise’ companies that treat them unfairly. But more positively, customers may learn to use information to negotiate for better terms (e.g., challenging a price offering with a robotic or live agent, searching deliberately for similar alternative products). Another issue concerns the skills of users of AI-enabled applications. Customers of diverse backgrounds, including those less skilled, should be able to benefit from consulting virtual smart assistants in performing simple tasks (e.g., ordering a familiar product). Yet, for conducting more complicated tasks (e.g., finding the optimal product, averting attempts of marketers to lead or draw the customer in a particular direction, performing tasks that involve multiple steps), customers may need greater technical skills to make efficient use of the appropriate AI-enabled applications (e.g., choosing, installing, and operating). Libai et al. caution that customers with higher CLV may have a stronger bargaining power to negotiate with a company; additionally, higher-income consumers with stronger technological know-how would be more likely to obtain the desired skills. Hence, disparity may arise in the means that should be available to consumers to gain from AI applications.
Experiences of encounters and interactions with AI application can go in opposite directions. If Dan gets what he wanted and is satisfied with the ease and convenience of getting there, everything is fine and everyone is happy. But, if something goes wrong for Diana and she is not satisfied with an outcome, the experience can become frustrating: companies are getting less approachable for human contact, the user may get ‘caught in a process’, and no remedy seems available. Consumers may also get wary of misuse of their personal data: whether the data is being used by a company for secondary purposes (e.g., marketing) other than helping with a given task at hand, and moreover if the data is shared with disapproved third-party partners. In worse cases, sensitive information could get leaked inadvertently by the system to another person (e.g., Danielle had details of a private conversation with Amazon’s Alexa sent to a person in her address book, causing her to feel “invaded”), consumers may feel being under surveillance by cameras, or more vulnerable when they choose to delegate a smart assistant to make purchase decisions and take actions on their behalf. These cases raise rightful concerns about loss of personal control and demands of consumers for higher transparency on how their personal information is being utilised (Puntoni, Reczek, Giesler, & Botti ).
The capabilities of AI-CRM can produce benefits for the firm that employs the technologies as well as for its customers. A major strength of an AI-CRM system is in enabling the company to respond in an adaptive, flexible and agile way to needs, preferences and actions of the customers. However, the advantages of AI-CRM may be used by companies to prioritise customers with higher customer lifetime value, leading to unfair and less desirable consequences. Care should be taken, therefore, to consider in every utilitisation and application of AI capabilities in CRM the value created for each party in the relationship: the company and the customer. It is doubtful if efforts to outsmart AI-enabled tools and agents to the benefit of the customer would come to any good. Consumers rather need to be open to opportunities AI can offer them (e.g., better-fitting products, convenience, faster response), yet remain vigilant to wrongful practices or treatment, and be ready to argue against and challenge unfair treatment or service in any interactive channels the company makes available to them.
Ron Ventura, Ph.D. (Marketing)
 Brave New World? On AI and the Management of Customer Relationships; Barak Libai, Yakov Bart, Sonja Gensler, Charles F. Hofacker, Andreas Kaplan, Kim Kotterheinrich, & Eike Benjamin Kroll, 2020; Journal of Interactive Marketing, 51 (August), pp. 44-56 (available for reading online at ScienceDirect.com)
 A Strategic Framework for Artificial Intelligence in Marketing; Ming-Hui Huang and Ronald T. Rust, 2021; Journal of the Academy of Marketing Science, 49, pp. 30-50 (available for reading online at link.Springer.com)
 Consumers and Artificial Intelligence: An Experiential Perspective; Stefano Puntoni, Rebecca Walker Reczek, Markus Giesler, & Simona Botti, 2021; Journal of Marketing, 85 (1), pp. 131-151 (available for reading online at journals.sagepub.com)