Editors Note: DBizInstitute is excited to share this article, written by Dr. Setrag Khoshafian, with our community and in advance of his new book release. Keep an eye on our website as we share additional articles in the coming months written by Setrag, as well as a pending Meet the Author webcast to discuss his new book ‘How to Alleviate Digital Transformation Debt’ expected to air Fall 2021. Be sure to pre-register for the webcast here! This article was originally published on CognitiveWorld.com on December 19, 2019.
There is no doubt about it, digital technology has become an essential tool for driving customer engagement. For many companies, effective digital solutions are the foundation on which positive B2C, B2B, and B2B2C relationships are founded. Technology fills in where legacy or manual systems could not, allowing businesses to create personalized, and intelligent, digital user-friendly journeys for a wide range of consumers.
AI is the most important digital transformation enabler in this new era of customer engagement
How much of customer engagement evolution is due to tech, and how is AI and other digital technologies-especially social, mobile, and connectivity with the Internet of Things-impacting the new generation of digitally transformed engagement solutions?
Difficult to surmise. But here are some observations.
The new generation of customers, especially Millennials, are increasingly tech-savvy and digitally chatty. With their mobile and connected devices, they are active and continuously engaged throughout social interactions. Above all, they are demanding experiences that are contextual, meaningful, instantaneous, and entertaining. Furthermore, social networking channels have given them a powerful voice–the voice of the network–allowing them to instantly provide feedback (good and bad) and share ideas about products, services, and companies. The traditional marketing and service approaches such as “throw it out there and see what sticks,” does not work–it actually has a negative impact for this new generation of consumers.
Hence the critical need for “intelligence” in customer engagements.
Customers express their feelings on products or services and these “sentiments” need to be listened to and intelligently analyzed– ignoring them could be detrimental to the product or services provider. Different Customers would like to be treated differently: through useful, meaningful interactions, offers, or decisions on their behalf that are specifically relevant to their context or situation.
AI in customer engagement can drive this differentiation
As their context or situation changes, so should the interactions, offers, or decisions. Furthermore, as they respond to offers, the underlying digital intelligence needs to learn from their behavior and the feedback needs to be used accordingly to adapt the customer offers or decisions. This is the very essence of AI for the digital customer engagement.
Example 1: This approach is yielding powerful results for organizations that have leveraged AI for customer engagement decisioning. Specific industries, such as telecoms, often struggle with “churn.” A large telecom provider using AI in customer engagement reduced church by 10%, increased sales revenue by 55% through intelligent customer service, and also modernized and simplified their agent’s environment through digitizing the operations.
AI: From Raw Data to Insight to Action
The modern digital consumer relies more on peer-reviewed “likes” as opposed to traditional marketing channels. The modern digital customer demands rapid responses–on time, on target, and customized to her or his constantly changing needs. Furthermore, these customers are using an increasing number of devices (a.k.a. IoT) that connect to their bodies (a.k.a. wearables), homes, businesses, cars, and more. With this steady uptick in digitally engaged consumers, data generation is exploding.
Customer Data can be historic or real-time (transactional). It can be structured, as in relational tables, or unstructured, as in text, audio, or video. It can be real-time event based or aggregated over time. Customer Data comes from multiple applications, social media or devices. Customer purchase patterns, customer sentiments expressed in social media, and customer transactions are some of the sources of data pertaining to the connected digital customer. In other words, the “Big Data” of customers.
Big data is characterized by large volume, variety, and changes with increasing velocity. The question is: Are we gaining insight or knowledge from the data that we are constantly generating? Customer satisfaction drivers and future behavior are all “hidden” in this data. So the more important question is, are we operationalizing this insight in intelligent customer interactions?
That is the essence of AI for customer engagement. Some of the top pragmatic applications of AI for customer engagements include “Cross-sell/Up-sell” of products and services. Other applications include “Campaign management,” “Customer acquisition,” and “Retention”.
Example 2: A large bank realized customer data is a gold mine so they used AI to create on spot targeted marketing campaigns that resulted in reduced marketing costs, improvement in customer satisfaction, and a 25% increase in responses.
Let’s touch upon some of the AI approaches and techniques that can be used for optimizing the customer engagement.
Predictive and Machine Learning
Prediction is ubiquitous. Almost every business flow or rule has some element of prediction to it. Most of the time, requirements arise from intuition, history, experience, or ad-hoc mechanisms to capture policies and procedures. Sometimes the original reasons for enacting these policies have long been obsolete. In contrast, predictive modeling is a scientific discipline within data mining that uses measurable predictors to predict the behavior of customers. These can be an ordinal or numerical value that can be predicted from other variable values. Historical data is analyzed and modeled to predict future behavior. Examples of predictors include purchasing preferences, geographical location, age, income, and properties pertaining to the history of activities. Predictive models can be discovered from either operational data, social interaction data, device data, or data warehouses.
There are many different types of predictive models including classification models, regression models, and clustering models, to name a few. In addition to aggregating and mining models from heterogeneous data, another benefit of AI is to opt for a system of continuous learning within the system itself. They need to know when the historic data used for modeling is no longer representative of current circumstances. With traditional predictive analytics, once the predictive model has been inferred from the data, it will not change anymore. In that case, the model will get “tired” and needs to be replaced by a new model based with more recent data-the model is derived from a snapshot of the data and immutable afterwards. Any organization that is responsibly using static predictive models will want to ensure that those models are continuously monitored.
Enter self-learning (a.k.a. machine learning or adaptive) approaches. Instead of looking at a snapshot of data, this model looks at a moving window of data as it enters the self-learning adaptive system. It’s still about predictability but now it is much more dynamic with continuous adjustments as sentiments, customer interactions, or customer device behaviors change. A popular use is in the digital customer marketing space where the decision strategy adapts to changes in customer behavior or market dynamics. Customer behavior can change because of demographic trends, legislation, interest rates, or a myriad of other factors such as how the customer is using their wearables or connected devices.
Static or more dynamic self-learning systems can leverage either a single source of data (e.g. customer inbound transactions) or more importantly an aggregation or fusion of data, multiple sources of data capturing customer behaviors, sentiments, or activities. The latter provides better assessment and opportunities for optimized customer engagements. The whole purpose of, and motivation for, analytics–static predictive or self-learning–is to discover these patterns (predictive models), use them to predict future behavior, and then act on the insight.
In Part II we shall delve deeper into the intelligent automation of the “act.” We will cover digitization of value streams, virtual assistants (bots), robotic process automation (RPA), AI assisted work automation, as well as real-time event processing for optimized customer experiences. The focus will be on executing the insights.
The term “AI” has evolved over the years and there are a number of academic and populist definitions. In this context, AI includes business rules, predictive, and machine learning of all sophistication and depth.
This is a two-part article series, be sure to check out Part II next!
