Source – cmswire.com
As machine learning algorithms are increasingly embedded in today’s analytics applications, this offspring of artificial intelligence is grabbing headlines.
In reality, however, the technology isn’t as new as you might think. Its roots in academia date back to the 1950s. Machine learning algorithms have risen out of predictive analytics and use data to learn and improve behaviors and recognize patterns — without being explicitly programmed to do so. While the technology may not be new, its business application are.
To better understand the impact of machine learning, let’s look how it intersects with customer experience and where this combination may be headed.
How Machine Learning Works
Machine Learning relies on automation and analytical algorithms to detect patterns and derive insight from data to achieve specific goals.
Until recently, machine learning was used only by organizations that had large analytical teams. Now machine learning is becoming democratized and, as a result, is not only impacting every day analytical programs (e.g., analytical segmentation, modeling and optimization,) it is also being used to achieve specific analytical goals.
These goals, particularly when they support improving customer experience, translate into personalizing (or even individualizing) an offer for a product or service. Personalization involves increasing the speed, accuracy and context of interaction. For example, imagine a father who receives a stellar offer for a pair of noise-cancelling headphones on his mobile device shortly after purchasing an electric guitar and amplifier. That’s an offer that has the right speed (right after purchase), accuracy (mobile as his preferred channel) and context (knowing he may need the headphones to silence the noise coming from his teenage son’s room). This type of offer increases customer satisfaction as it is pertinent at the time of need. The end-result is a happier customer (this company understands me) and a happier brand (increased revenues.)
So, what are some practical applications we are seeing today? Below are a few examples of how machine learning is improving customer experience through next-best-offer personalization, customer behavior analytics with new data sources and analytical optimization. The examples are mainly from financial services organizations (banks and insurers), but similar use cases and results are being seen across other industries as well.
Machine learning, particularly predictive models, helps banks personalize offers with an extreme amount of certainty. The results:
Customers interacting via branch and contact center channels do not have to negotiate nearly as much to receive an offer they want.
Certain segments of banking customers are receiving tailored offers initially instead of irrelevant non-contextual offers that have a low degree of probability for conversion.
Data and analytics allow regional banks to understand customer preferences — including preferred rates and terms — making the setup process more efficient for both bank and customer.
Customer Behavior Analytics with New Data Sources
Using machine learning in combination with new data sources, whether it is Internet of Things (IoT,) telematics, geographic or social data is leading to an augmentation of the customer profile and an increased understanding of how and why an insurance customer behaves in the manner that they do. The results:
Property and casualty insurers can use vehicle data (pending opt-in) to better understand the best policies to offer drivers. No longer does the 16-to-25-year-old male automatically get the highest premium.
Health and life insurers are collecting device data from fitness devices (pending opt-in) and combining it with machine learning algorithms to better segment and classify customers for certain fitness or wellness programs.
Optimization of this nature is not priority or business-rule based, but is instead rooted in operations research and analytical algorithms, and helps financial services companies not only predict how a customer may behave or act, but also how the competition may move in the future. The results:
Financial services companies are using machine learning algorithms (random forest and gradient boosted models) to predict the probability to be ranked in a certain place (i.e., top 3) in an aggregator portal for financial services products (policies, bank cards, loan rates and so on).
Insurance companies are using random forest and gradient boosted trees to predict mid-term cancellation rates on policies.
Banks are able to predict volumes for credit card lines and adjust rates and terms accordingly in order to attract the right type and volume of customers for a certain product.
These are the practical applications we are seeing today, mainly where machine learning operates behind the scenes — as an accuracy, speed, and efficiency enabler. But what may happen in both the near- and long-term with regard to the intersection of machine learning and customer experience? Three main trends will emerge:
1. The Use of Augmented Analytics
Augmented analytics combines data preparation, business intelligence, predictive analytics and machine learning capabilities into a single, more automatic process than we have seen to date. In the near future, augmented analytics capabilities will help organizations prepare and cleanse data, find key insights and hidden patterns, and report on findings in an automated fashion. Data volumes of today have created millions of variable combinations that are nearly impossible to process manually, thus automation via augmented analytics will be needed. The result will be quicker time to insight, which will affect the customer experience through continual improvement. Customer frustration levels will decrease and delight will be on the rise. Rita Sallam of Gartner has co-authored a great article on the topic here.
2. Collaborative Machine Learning Replaces Collaborative Filtering
Many open source machine learning libraries, algorithms and frameworks will join forces to become stronger and thus replace some of the lower-level personalization we see today, namely the approach of collaborative filtering. Collaborative filtering is the technique most commonly used in recommender systems, think Amazon and Netflix. The result will be an even higher degree of personalization and contextualization than what we have seen to date, resulting in better music, food, movie, travel, product and purchase recommendations than ever before.
3. Neural Networks and Decision Trees to the Forefront
The techniques that are beginning to be used with machine learning, neural networks and decision trees will be more prevalent as use cases, which will require these techniques be incorporated into business operations. Neural networks will support better classification and forecasting, while decision trees will support more complex rule and relationship-based customer experience programs.
Machine Learning: The Future Impact
Machine learning will increase automation in industries like manufacturing and medicine, but from a customer experience perspective, we will see a continual embedding of machine learning into customer experience programs. Business process automation of these programs will continue to increase — from both an operational and executional perspective. This will improve the organization’s ability to support complex decisions, forecasts and optimizations.
As the organizational understanding of the customer continues to improve, the delivery of offers and interactions will be further contextualized. The beneficiary — customer experience programs and the metrics that track their success — customer lifetime value, net promoter score and so on. Based on the progression over just the past few years, I can’t wait to see how organizations will delight customers in the near future.