Read an Excerpt
Customer Experience Analytics
The Key to Real-Time, Adaptive Customer Relationships
By Arvind Sathi
MC PressCopyright © 2011 IBM
All rights reserved.
The Industry View
In this chapter, we introduce customer experience and Customer Experience Analytics (CEA) using a set of industry examples. Each industry has a different name for its customers — subscribers, citizens, patients, drivers, viewers, and so on. However, the examples will illustrate similar characteristics. In each case, customer data is collected and harnessed to create insights about the customers, using a set of predictive models. The enriched information is used to drive improvements in products or customer-facing processes. In each industry example, I have chosen examples that are personal and commonplace, so we can relate easily to them and realize the level of disruptive change they can bring to suppliers around us.
Customer Experience Analytics Through Examples
Davenport and Harris have defined analytics to mean "the extensive use of data, statistical and quantitative analysis, explanatory and predictive models, and fact-based management to derive decisions and actions." Let us use a set of industry examples to show how analytics as defined by Davenport and Harris is being applied to a number of consumer-facing industries. We encounter good as well as bad customer experience from a variety of suppliers. While each of us may have different measures for evaluating customer experience, we can easily spot good encounters based on the attention, personalization, trust, or emotional attachment a supplier provides to us. Some suppliers are far better at analyzing our experience and using that understanding to improve their product, customer service, price, location, et cetera. The experience is pervasive across a cross-section of industries (see Figure 1.1).
Let me take a couple of industries to narrate customer experiences and how analytics is used to organize, collate, mine, and improve customer experience. Using a set of case studies, I will show how customer data is organized, analyzed, and incorporated into business decisions.
Communication Service Providers
Let me start with our experience as communications customers. We all have often experienced issues related to call and Internet data quality. In June 2010 at Apple's Worldwide Developers Conference (WWDC), CEO Steve Jobs decided to demonstrate video conferencing capabilities on the iPhone. As he was connecting his iPhone to demonstrate the video calling features, the connection failed. Because this was Steve Jobs demonstrating a new cool feature, you can be sure that the hundreds of reporters and analysts sitting in the room were each connected to the facility's Wi-Fi and trying to upload the news and related information to their respective sites. Unfortunately, this crowding of devices took a toll on the device Steve Jobs was using. As he tried repeatedly to connect his iPhone, the connection kept failing. It is important to note that in this case the problem causing the dropped calls was right in the room, not with a communications service provider, a device, or an application running on a device. The video shows the problem getting fixed as Steve Jobs repeatedly tells the audience to shut down their devices.
The IT support organization deserves a big accolade for identifying the problem with the failed connections in near real-time. Steve Jobs was persistent in fixing the problem. After trying to connect a couple more times, he asked the audience to disconnect their devices so he could proceed with the demo. You can watch a video clip of his attempts to fix the problem on YouTube at http://www.youtube.com/watch?v=RGVsGSimLJg.
Although this particular example was not caused by the service provider, our gut reaction would be to be to blame the service provider. I asked some of the communications service providers how they keep track of their premium and VIP customers, how they excel in their service to their most important customers, and how they manage their customer perception for premium customers. Traditionally, Service Level Agreements (SLAs) were managed through service contracts. Most wireless providers traditionally offered loyalty points or service discounts if they faced service problems. For mass markets, that may be a good start. However, as the Steve Jobs video demonstrates, this case involved a very important wireless device, whose failure would be openly discussed by analysts and posted on YouTube.
How do we know the calls were dropped? The information is available to a service provider. Are they analyzing this information to trigger corrective actions before the service reaches a level of intolerance? As we see in the video, Steve Jobs made repeated attempts during the demonstration. Could we count five dropped calls and trigger a response once we reach that threshold, especially when dealing with premium customers? How do we know if five dropped calls is a normal threshold for customer churn? How do we identify the service performance to the phone owner? How do we know that the service interruption is impacting a premium customer? The cause of a dropped call could be anything, whether it is the service provider network, the roaming network, or the device.
Clearly, service providers cannot monitor every subscriber and provide personalized attention. Is it possible for us to differentiate service for the best customers? How do we prioritize network operations to deal with premium customers first? The information about the service interruption is being generated in real-time. Can we collect this information, correlate with premium customers, analyze to establish root cause, and work with the subscriber to fix the problem? Can we work on this in near real-time so the subscriber may start receiving better performance right away? Can we involve the sales teams, the customer service organization, the touch points, and the device, so all of them are aware of the problem and are helping the subscriber in solving the problem?
I often work from my home office if I am not traveling to client sites, and I often need to download large documents. I decided on a 10 Mbps DSL line to reduce the download time on these large files. Unfortunately, much to my dismay, downloads remained painfully slow. I was on the verge of calling my cable company to experiment with cable modems. Unexpectedly, I received a call from my telecommunications service provider. The caller introduced himself as a network operations person and told me he was concerned that I was getting only 10 percent performance on my DSL line. He asked me if I had any analog devices connected to the phone line without a DSL filter. After surveying the house, I found that my home security alarm system was connected to the phone line directly, without the DSL filter. He told me that I needed to contact my home security provider for a special filter.
After a little bit of discussion, I decided to go for a second line instead, so that I could dedicate a line to my home office. I was promptly transferred to the customer service representative (CSR). In my discussion with the CSR, I discovered that I was using a metered long-distance product that offered me monthly calls up to a certain number of minutes. At the same time, I was hitting the upper limit each month due to my business calls from the home phone. By moving to an unlimited plan, I reduced the communications service provider's task of reporting a log of my long-distance calls, and I saved myself $10 per month. Eventually, I decided to apply the $10 toward getting a 20 Mbps DSL service.
What happened here? I had a sub-optimal telephone configuration for my needs, and I was a perfect candidate to churn. The call from network operations not only saved me as a customer but also increased the revenue through a second line and made the service far closer to my requirements. The service I received was very personalized. The call was totally unsolicited but based on specific performance data from my line. My discussion with the CSR was equally interesting. Instead of just selling a second line to me, he used information about my usage to compare products for me and offer me one that saved me money and was more useful.
As competition and cost pressures have intensified, corporations worldwide are on a mission to improve customer service and reduce operational costs using technology. Unfortunately, for a number of years, the pendulum kept shifting toward cost reduction. Process consultants clocked every second of call center time and every transaction in customer service, looking for ways to reduce sales costs. The end result was multiple hierarchies of IVR trees, impersonal call centers, and rigorous implementation of organizational policies — ending in poor customer experience.
My experience above is very contrary to this cost-cutting trend. The customer-facing personnel used precious time to patiently work with me on my requirements, finding problems with my installation and the billing options currently on my plan. They offered me recommendations that improved my service and their future prospects for keeping me as a long-time customer. In doing so, they relied heavily on information about me known to them — products purchased and their subsequent usage. Also, they were supported by their data warehouse and analytics applications, which pinpointed to them that they were at risk of losing me, a premium customer, due to the trickle feed I was getting on my bandwidth. They were able to focus on customers with high revenue and the highest risk of defection.
Let me now move to the financial services industry. For my house, I had a home mortgage as well as a home equity line. While both loans came from the same mortgage provider, they were sold to me separately, and I ended up setting up online access to both accounts. I had two different user names and passwords, and each time I accessed my account to make changes, such as cell phone contact information, I had to repeat the changes in both accounts. One fine morning, I logged into my home mortgage account and found that the home equity account was now showing up along with the home mortgage account in a single login.
My bank had used a Master Data Management (MDM) program to consolidate mortgage accounts. My account was a perfect candidate to be discovered in the "harmonization process." Both accounts had the same name, social security number, phone number, and address. The only task the bank had was to seek my authorization, and now I was relieved from double typing changes to both accounts.
At the height of the real estate bubble, we all excitedly purchased houses, looking for the lowest mortgage rates. Banks competed bravely, converting mortgages into commodities to be traded as securities. What happened to the customer experience? Mortgage shopping for the lowest interest rate resulted in multiple accounts.
Mortgage companies have limited visibility to our overall portfolio. Between my house and rental property, I have four mortgage loans with three different mortgage providers. In addition, I have two banks helping me with my cash accounts. Any time a change occurs in my personal information, I have to enter it in five places. I am the ultimate information integrator. Each of these five providers has a very small window of visibility to my portfolio and, hence, no ability to offer any consolidations. At the same time, an enormous opportunity exists for any one of them to increase their share of the wallet, if they could show me an integrated front. This happened to my bank; they discovered I had two loans with them, and I found it easier to deal with them, which should positively impact my future mortgage company selection in their favor.
As I studied the problem, I discovered yet another reason for my bank to consolidate my accounts. Banks are very concerned about the overall risk associated with a customer. As we take mortgages, use their investment arms for margin calls, and take auto and home equity loans, we are increasingly making the bank vulnerable to household risk of failure. If the housing market slides and leads to a foreclosure on the house as well as a bankruptcy, all the accounts related to the individual are impacted. The cleanup begins with the accounts held with the same bank. If using a harmonization process, the bank can discover all the accounts the individual has with the bank, and the risk can then be computed at the household level.
By analyzing social media data, we can discover household relationships, as well as business relationships across individuals. This data can be used for far more accurate computation of overall risk. This information can also be used to identify fraudulent behaviors or money laundering, and for other risk-related identity applications.
Banks can combine their data with outside data to gain an accurate understanding of the customer. Credit-card operations for a bank are good candidates for this total real-time view of the customer. Credit-card companies have invested significantly in fraud-detection systems. These systems operate in near real-time and evaluate each transaction made by a customer using the customer's past historical purchase record — time of purchase, location, type of purchase, and transaction value, to name a few. Any deviations from the past are examined closely and accepted or rejected based on rules accumulated over years of fraud detection. The solution has a real-time analytics component, which provides the capability to evaluate a transaction while the transaction is being processed. It also has a predictive modeling component that creates fraud-detection rules based on analysis of historical data. Credit-card operators constantly monitor the effectiveness of these rules and modify them based on their success in detecting fraud while maintaining a good customer experience in the use of credit cards. If the fraud detection resulted in questioning of every transaction, it would drive up the execution costs and imperil customer experience. If fraud detection resulted in the passage of fraudulent transactions, it would drive losses to the credit-card operators and customers. The optimization involves a fast realtime analytics engine and an ability to use predictive modeling to further modify this engine.
My work requires me to travel frequently, almost once a week. While traveling, I use my corporate card all the time and use my personal credit card sparingly — using it only when I need to make any personal purchases. This behavior poses a problem to the real-time analytics and monitoring of my personal card because the usage is sporadic and geographically diverse in a random manner. Invariably, the credit card is denied at the time of purchase, requiring me to phone the call center for security verification. These calls are expensive. I remember talking to a support line three times from India, with each call taking ten minutes or longer. The overall cost of the call, including telephone charges, the call center agent's time, and my time, adds up.
While I am thankful to the credit-card company for taking my card security seriously, I was curious whether there was an easier way for them to deal with me. I asked the credit-card call center agent how I could make the card company's monitoring easier, and the response was to call them before each trip. This solution might reduce the number of times my credit card is denied; however, it would significantly increase the call-center costs. And I would have to call each time before traveling, which could be a lot more than the number of times I use my personal credit card, or the times credit is denied.
The premise for credit-card fraud is that someone could steal my credit card and use it. A typical fraud rule looks for an unusual purchase initiated in an international location. Unfortunately, for frequent travelers like me, regular credit-card use can easily mimic these fraudulent transactions. As I travel to Mexico frequently but use my personal credit card rarely, each of those purchases is very likely to be tagged as unusual activity. However, I carry a credit card and a smart phone all the time when I travel. Although my credit-card company may not know of my travel to distant geographies, my smart phone has full awareness of my location. Also, the chances of my losing both my credit card and my phone are significantly less. If only I could authorize my credit-card company to check my phone location each time there is a concern about the credit-card usage, and even place an app on my phone to ask me to authorize the charges using a secure login or password to eliminate possibility of my phone being stolen at the same time.
Financial institutions are rapidly discovering their partnership with the phone companies. Today, Chase offers mobile check deposit using the Apple iPhone. Using the camera in an iPhone, I can take a picture of both sides of the check and then use the Chase Mobile app on my iPhone to log into my account with a special authorization ID supplied by Chase. Now that my phone and bank are aware of each other, they can use this information for a variety of applications to improve my customer experience.
The risk and credit-card fraud analytics covered above for the financial services has another set of buyers: federal, state, and local government, who analyze money laundering, tax fraud, and related criminal activities. The losses due to fraud are enormous. These losses can stem from tax evasion, misappropriation of funds, or fraudulent benefit applications. CEA can provide public services with the tools to enable fraud analytics using predictive models and the enormous social media and third-party data available to them
Over the past couple of years, we have seen a rapid growth in misuse of technology for organizing "flash mobs." In a typical flash mob, the suspects often connect via cell phones or social media sites and converge on a spot to engage in stealing goods from stores or assaulting bystanders. USA Today reported on one such incident in August: "Philadelphia leaders imposed an early curfew on parts of the city this month after roving bands of teens beat and robbed bystanders during violent attacks across the city. Surveillance cameras caught several dozen youths swarming into convenience stores in Germantown, Maryland, and Washington, D.C., and stealing armfuls of snacks and drinks as the store clerk looked on helplessly."
Excerpted from Customer Experience Analytics by Arvind Sathi. Copyright © 2011 IBM. Excerpted by permission of MC Press.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.