Chapter 22
Improving Business Processes with Big Data Analytics: A Real-World View
In This Chapter
Seeing why companies need big data analytics
Improving customer service with text analytics
Predicting the next best action with big data
Preventing fraud with big data
Understanding the business benefit of integrating new sources of data
It is becoming clearer every day that business decision makers need to be able to interpret data differently if businesses expect to keep up with rapid market changes. C-level executives know that future success depends on innovation and building more predictive, responsive, and personalized customer experiences. Success also depends on reducing risk and making governance and security a priority. Meeting these changing business requirements demands that the right information be available at the right time. Many companies see big data analytics as central to their business strategy to increase the level of partner and customer engagement and to decrease the time to decision.
In this chapter, you find out about organizations and entire industries that are changing the way they manage and analyze structured and unstructured data that is increasing in volume, velocity, variety, and veracity. We show how the technology presented in previous chapters can be applied to solve real-world business problems.
Understanding Companies’ Needs for Big Data Analytics
The data that can make a difference in how companies satisfy their customers and partners is not necessarily in traditional databases any more. The value of unstructured data from nontraditional sources has become apparent. Business leaders have discovered that if they can quickly analyze information that is unstructured — either in the form of text from customer support systems or social media sites — they can gain important insights. When companies can analyze massive collections of data and compare those results in real time to the customer decision-making process, businesses can gain huge revenue increases. Therefore, leveraging a combination of unstructured and structured data as part of a business process can transform a business’s capability to be agile and nimble, and most importantly, profitable.
Improving the Customer Experience with Text Analytics
Many companies accumulate huge amounts of unstructured data that have been underutilized as sources of information about their customer experience. Unstructured data is the text found in e-mails, text messages, call center notes, comments in survey responses, tweets, and blogs. This type of data represents about 80 percent of the data available to companies, and it is continuing to grow. Unstructured data has typically taken many manual hours to review, and in many companies, it has never have been adequately analyzed. Companies recognize that if this data is analyzed at the right time, it may help to identify patterns of customer dissatisfaction or a potential product defect so that corrective action can be taken before it is too late. The increasing sophistication of text analytics is viewed by companies as a major benefit, enabling the deep analysis of large volumes of unstructured data in real or near real time so that the results can be used in decision making. Text analytics is the process of analyzing unstructured text, extracting relevant information, and transforming it into structured information that can be leveraged in various ways. (Text analytics is covered in more detail in Chapter 13.)
How would this work in the real world? Look at an example of a car rental company that was experiencing huge pressures from emerging companies that didn’t have the same high overhead. How could the existing company compete? Improving responsiveness seemed to be the key to success. Therefore, the company was able to use text analytics to begin making significant improvements in its customer service. The company encouraged its customers to provide feedback on its services in online surveys or by e-mail or text. Customers used these communication methods to provide comments about service issues such as longer-than-expected wait times, poor agent service, or not getting the car they ordered. However, the company’s response and interpretation of these comments had been inconsistent. The company was taking the right approach, but the response was too slow and the analysis was inconsistent. Agency managers read the e-mails and comments in web surveys and text messages. Managers read the comments online and placed them in categories for future attention. Unfortunately, this approach took a long time, and each manager followed a different approach to categorizing comments. It was too easy to miss patterns of dissatisfaction or concern that might show up if you were able to look across a large number of comments at one time.
What managers wanted to do was analyze feedback from customers faster so that they could identify potential issues in real time and address problems at the outset before they become bigger problems. Managers implemented a text analytics solution that allowed them to quickly analyze text for insight across all types of sources, including structured and unstructured data. They also implemented a sentiment analysis solution that enabled an automated approach to identifying forms of communication that might need immediate attention. They were able to capture large volumes of information about the customer experience in real time and quickly analyze and take action.
The business value to the big data analytics implementation
The company was able to make major improvements in customer satisfaction. It is able to keep better track of car and equipment rental performance levels and find problems and fix them early. It now has a more accurate understanding of where problems are located and can recognize them much faster. The new analysis provided managers with an early identification of problems at one location. As a result, they were able to make changes and improve customer satisfaction at this location.
Using Big Data Analytics to Determine Next Best Action
Today the customer is in the driver’s seat when it comes to making a choice about how to interact with a service provider. This is true across many industries, including telcos, insurance companies, banks, and retailers. The buyer has many more channel options and is increasingly researching purchase decisions and making buying decisions from a mobile device. You need to manage your customer interactions armed with in-depth and customized knowledge about each individual customer to compete in a fast-paced, mobile-driven market. What does it take to provide the right offer to a buyer while he is making a purchasing decision? How do you ensure that your customer service representatives are armed with customized knowledge about your customer’s value to the company and her specific requirements? How can you integrate and analyze multiple sources of structured and unstructured information so that you can offer customers the most appropriate action at the time of engagement? How do you quickly assess the value of a customer and determine what sort of offer that customer needs so that you can keep the customer satisfied and make a sale?
Company executives are increasingly viewing big data analytics as the secret weapon they need to take the next best action in highly competitive environments. Using analytics to understand customer requirements on a more personal level is seen as an important capability when dealing with the increased pressures of an empowered consumer. Companies are expanding their use of social media and mobile computing environments and want to reach their customers at the right time through their channel of choice. To deliver successful customer outcomes in a mobile world, offers need to be as targeted and personal as possible. Companies are using their analytics platform combined with big data analysis with fast processing of real-time data to achieve competitive advantage. Some of the key goals they wish to achieve include
Increase their understanding of each customer’s unique needs. Provide these in-depth customer insights at the right time to make them actionable.
Improve responsiveness to customers at the point of interaction.
Integrate real-time purchase data with large volumes of historical purchase data and other sources of data to make a targeted recommendation at the point of sale.
Provide customer service representatives with the knowledge to recommend the next best action for the customer.
Improve customer satisfaction and customer retention.
Deliver the right offer so that it is most likely to be accepted by the customer.
What does a next best action solution look like? Companies are integrating and analyzing large volumes of unstructured and streaming data from e-mails, text messages, call center notes, online surveys, voice recordings, GPS units, and social media. In some situations, companies are able to find new uses for data that was too large, too fast, or of the wrong structure to be incorporated into analytics and predictive models before. The models that companies are able to build are more advanced and can incorporate real-time data from a variety of sources. Company analysts are looking for patterns in the data that will provide additional insight into customer opinions and behavior. Speed is a top priority. Your model needs to predict the next best action very quickly if you want to be successful in this fast-paced mobile world.
Advanced technology is helping companies to generate actionable information in minutes instead of days or weeks. Predicting the next best action often requires the use of sophisticated machine-learning algorithms from a cognitive computing environment like IBM’s Watson. Watson can be used to process large volumes of data and analyze data in motion and to understand customers and present an immediate response, using cognitive computing to personalize offers. (Refer to Chapter 13 for information on Watson.)
We look at several real-world examples of companies in the financial services industry that are investing heavily in new ways to understand and respond to customers.
An insurance company wants to increase the efficiency and effectiveness of its call center representatives. Agents could not quickly identify the full extent of a customer’s mix of business, and therefore, it was hard to identify top customers who needed special attention. In addition, agents found it very time-consuming to search for call notes captured during previous calls with a particular customer and found that information that might have helped to solve the problem was not located until it was too late. Unfortunately, a growing number of customer interactions resulted in dissatisfied customers. This company implements a solution that transforms conversations from recorded calls into text. Keywords are identified and analyzed. This data is combined with historical data about the customer to identify high-priority customers who require immediate attention and to deliver a more timely and appropriate response to all customers.
A global bank is concerned about the length of time it takes to access customer information. It wants to provide call center representatives with more information about customers and to have a better understanding of the network of customer relationships (family, business, and social networks). Executives have large volumes of structured and unstructured information about customers, including e-mails, letters, call center notes, chats, and voice recordings. The bank implemented a big data analytics solution that improves the way its representatives support customers by providing them with an early indication of each customer’s needs before they got on the phone. The platform uses social media data to understand relationships and can determine whom the customer connected to. The solution combines multiple sources of data, both internal and external. Some indication may exist of major life events that are taking place for this customer. As a result, agents are able to take the next best action. For example, a customer may have a child ready to graduate from high school, and this might be a good time to discuss a college loan.
A credit card company wants to increase its capability to monitor customer experience and take action based on each customer’s unique situation. It wants to tailor its solution to the individual customer and not to a demographic. As a response, the company developed a big data analytics solution that integrates information from traditional structured sources, such as customer transaction information, with unstructured and streaming data such as click-stream data, Twitter feeds, and other social media data. Its immediate goal is to create detailed microsegmentations of customers to be able to provide targeted offers. The solution provides the company with an effective approach to analyzing lots of information quickly to identify customer intent to buy and create a personalized next best offer for that customer.
Preventing Fraud with Big Data Analytics
By many estimates, at least 10 percent of insurance company payments are for fraudulent claims, and the global sum of these fraudulent payments amounts to billions or possibly trillions of dollars. While insurance fraud is not a new problem, the severity of the problem is increasing and perpetrators of insurance fraud are becoming increasingly sophisticated. Fraud occurs in all lines of the insurance business, including automobile, health, workers’ compensation, disability, and business insurance. Fraud may be committed by an individual who falsifies a claim of a broken arm after staging a fall in a shopping mall or by any number of business workers who have some association with the process of repairing damage from accidents, treating medical injuries, or dealing with other aspects of the claims process. The practice of insurance fraud is widespread and may include organized crime groups involved in car repairs, medical treatment, legal work, home repairs, or other functions related to the claim.
What is the role for big data analytics in helping insurance companies find ways to detect fraud? Insurance companies want to stop fraud early before they get involved in the processing of the claim. By developing predictive models based on both historical and real-time data on wages, medical claims, attorney costs, demographics, weather data, call center notes, and voice recordings, companies are in a better position to identify suspected fraudulent claims in the early stages of interaction. For example, a personal injury claim could potentially include fake medical claims or a staged accident. Companies have seen an increase in sophisticated crime rings to perpetrate auto insurance or medical fraud. These rings may have similar methods of operation that are enacted in different regions of the country or using different aliases for the claimants. Big data analysis can quickly look for patterns in historical claims and identify similarities or bring up questions in a new claim before the process gets too far along.
Risk and fraud experts at insurance companies, along with actuarial and underwriting executives and insurance business managers, all see big data analytics as having the potential to deliver a huge benefit by helping to anticipate and decrease attempted fraud. The goal is to identify fraudulent claims at the first notice of loss — at the first point where you need an underwriter or actuary.
Consider the following real-world example. An insurance company wants to improve its ability to make real-time decisions when deciding how to process a new claim. The company’s cost outlay including litigation payments related to fraudulent claims has been rising steadily. The company has extensive policies in place to help underwriters evaluate the legitimacy of claims, but the underwriters often did not have the data they needed at the right time to make an informed decision. The company implemented a big data analytics platform to provide the integration and analysis of data from multiple sources. The platform incorporates extensive use of social media data and streaming data to help provide a real-time view. Call center agents are able to have a much deeper insight into possible patterns of behavior and relationships between other claimants and service providers when a call first comes in.
For example, an agent may receive an alert about a new claim that indicates the claimant was a previous witness on a similar claim six months ago. After uncovering other unusual patterns of behavior and presenting this information to the claimant, the claim process may be halted before it really gets going. In other situations, social media data may indicate that conditions described in a claim did not take place on the day in question. For example, a claimant indicated that his car was totaled in a flood, but documentation from social media showed that the car had actually been in another city on the day the flood occurred.
Insurance fraud is such a huge cost for companies that executives are moving quickly to incorporate big data analytics and other advanced technology to address the problem of insurance fraud. Insurance companies not only feel the impact of these high costs, but the costs also have a negative impact on customers who are charged higher rates to account for the losses. By using big data analytics to look for patterns of fraudulent behavior in enormous amounts of unstructured and structured claims-related data, companies are detecting fraud in real time. The return on investment for these companies can be huge. They are able to analyze complex information and accident scenarios in minutes as compared to days or months before implementing a big data platform.
The Business Benefit of Integrating New Sources of Data
Big data analytics is providing companies with a new way to provide answers to some age-old questions. Businesses have traditionally focused on how to improve customer service, provide the right offer to the right customer at the right time, and reduce risk and fraud. So what’s changed? By integrating new sources of unstructured data such as web logs, call center notes, e-mails, log data, and geospatial data with traditional sources of transaction, customer, and operational data, companies can look at their businesses much differently. They can gather data they were not able to collect previously and use this data to look for patterns of behavior that provide a great insight to the business. Integrating all these sources of data provides a way for companies to deepen their understanding of customers, products, and risk.