IBM Business Analytics Case Studies

IBM Global Center for Smarter Analytics

Fundamentals of Business Analytics

Case Study


2

________________________________________________________________________________

Licensed Materials - Property of IBM © Copyright IBM Corporation 2013 Published May 2013

IBM, the IBM logo and ibm.com are trademarks of International Business Machines Corp., registered in many jurisdictions worldwide. Other product and service names might be trademarks of IBM or other companies.

This case study set contains proprietary information which is protected by copyright. No part of this document may be modified without a legal license agreement from IBM Corporation. Any references in this information to non-IBM Web sites are provided for convenience only and do not in any manner serve as an endorsement of those Web sites. The materials at those Web sites are not part of the materials for this IBM product and use of those Web sites is at your own risk.

IBM Global Center for Smarter Analytics Case Analysis 1: Cincinnati ZooCase Analysis 1: Cincinnati ZooCase Analysis 1: Cincinnati ZooCase Analysis 1: Cincinnati Zoo

IBM Global Center for IBM Global Center for IBM Global Center for IBM Global Center for Smarter AnalyticsSmarter AnalyticsSmarter AnalyticsSmarter Analytics

ComComComCompany Backgroundpany Backgroundpany Backgroundpany Background

The Cincinnati Zoo & Botanical Garden is one of the most popular attractions and a Top Zoo for Children according to Parent’s Magazine. Each year, more than 1.3 million people visit its 71-acre site, which is home to more than 500 animal and 3,000 plant species. Although the Zoo is a non-profit organization and is partially subsidized by Hamilton County, more than two thirds of its $26 million annual budget is generated through its own fundraising efforts.

Cincinnati Zoo takes pride in the fact that it has the lowest public subsidy of any zoo in Ohio and generates more than two thirds of its $26 million annual budget through its own fundraising efforts. In challenging economic conditions, the Zoo wanted to reduce its reliance on subsidies even further by increasing visitor attendance and revenues from secondary sources such as membership, food and retail outlets. This would secure the Zoo’s future and enable it to continue its pioneering work in conservation, preservation and research.

Currently, Cincinnati Zoo makes use of four separate

legacy point-of-sale systems with a single platform to provide data on all admission, membership, retail and food service sales. To monitor the number of visitors in certain areas, the zoo staff members make their rounds to observe.

You have been hired by Cincinnati Zoo to provide them a solution that will enable them achieve their business objectives.

BusinessBusinessBusinessBusiness ObjectivesObjectivesObjectivesObjectives

• Increase attendance and revenues by enhancing the customer experience for each visitor

• Boost sales for food and retail outlets through more effective marketing and promotions

• Optimize labor costs by gaining a better understanding of demand patterns throughout the year

• Maximize the workforce by identifying areas of the park that are in need of more staff

QuestionsQuestionsQuestionsQuestions

1. What are the problems with the existing system of the Cincinnati Zoo?

2. What are the data that are available to the zoo management and staff?

3. How can these data be used to help the following perform their work better to achieve the business objectives of the Cincinnati Zoo?

a. Zoo administrators b. Zoo staff

4. What solutions can be provided for the following stakeholders so that the business objectives of Cincinnati Zoo are achieved? Explain in detail the proposed solution for each stakeholder. In each solution, identify which type of analytics will be applied (i.e., descriptive, predictive or prescriptive).

a. Zoo visitors b. Zoo management c. Zoo staff members

_________________________________________

© Copyright IBM Corporation 2013 IBM Philippines IBM Plaza, Libis, Quezon City May 2013


4

Case Analysis 2: Kraft AustraliaCase Analysis 2: Kraft AustraliaCase Analysis 2: Kraft AustraliaCase Analysis 2: Kraft Australia

Company BackgroundCompany BackgroundCompany BackgroundCompany Background

Kraft Foods Australia is a subsidiary of Kraft Foods, the second largest branded food and beverage company in the world. Established in 1926, Kraft Australia is headquartered in Melbourne and has sales revenue of over A$650 million. The company’s flagship brand—Vegemite—has long been considered an Australian national icon.

For all the diversity in the worldwide consumer products market, the most successful companies tend to have an important thing in common. They focus on continually improving the business functions that are critical to the consumer products business model, from maintaining vibrant and market-driven product development to having efficient manufacturing and a lean and flexible supply chain. Even more fundamental to consumer-products success, however, is the strength of the relationships companies form with their customers. That’s why a company’s brands, as the embodiment of this bond, represent the most important strategic asset for a consumer products company—indeed the backbone of its success. In a business built on customer loyalty, brands also serve to convey the promise of a consistent and high-quality experience that customers can count on. Maintaining the integrity of this promise—and thus the strength of the brand—is a first level priority for consumer products companies and a key to maintaining and growing market share.

The billions of dollars companies spend every year on brand research and advertising attests to the continuous nature of this challenge. With demographics constantly changing and new market opportunities opening up, consumer products companies need to not only strengthen their brands,

but also ensure that they are in sync with marketplace trends. It was through such an exercise that Kraft Australia (www.kraftfoods.com.au) made an important discovery. Its hero brand, Vegemite, has been a beloved fixture in Australian households since it was introduced half a century ago. While its brand remained strong, Kraft found that a large portion of “new” Australians—such as those who immigrated into Australia—had no relationship with the Vegemite brand. Seeking to tap into this market potential, Kraft was determined to gain a comprehensive understanding of this group, and more specifically its beliefs and attitudes toward Vegemite. At the same time, Kraft was also looking to bring its longtime brand message—which featured children as “Happy Little Vegemites”—up to date with changing lifestyles, demographics and usage patterns. Kraft also understood that there were many lapsed users who had grown up on the spread who needed to be reminded of their affinity for the brand. Kraft didn’t take such change lightly. It was determined to get the deepest possible insights into what consumers were thinking and saying about the Vegemite brand and to tailor a message that would resonate most strongly with those themes.

The web offers a wealth of the insights from consumers through web content. There are half a million mentions of Vegemite out of 1.5 billion posts of user generated content across 38 languages. These can provide Kraft the answer that they may need.

One of Kraft’s goals going into the research was to test certain hypotheses about the kind of issues it would address in its upcoming advertising campaign. An example was whether or not customers were looking for variations to the “classic” Vegemite product, and if so what they should be (e.g., new flavors, different jar sizes).

Business Objectives Business Objectives Business Objectives Business Objectives

• Change the branding campaign based on how the consumers view and used Vegemite

• Identify market opportunities at a very early stage

• Detect and respond to threats to Kraft’s brands and corporate reputation

• Ability to increase sales and customer loyalty through more targeted advertising campaigns


5


1. What are the problems that Kraft Australia would like to address?

2. What are the data available that can be used for an analytics solution?

3. How can these data be used to identify how the customers view and use Vegemite?

4. What are the possible solutions that can be provided for Kraft Australia to help them solve their problems?

_________________________________________

© Copyright IBM Corporation 2013 IBM Philippines IBM Plaza, Libis, Quezon City

May 2013

IBM Global Center for Smarter Analytics Case Analysis 3: Papa Gino’s, Case Analysis 3: Papa Gino’s, Case Analysis 3: Papa Gino’s, Case Analysis 3: Papa Gino’s, Inc.Inc.Inc.Inc.


Based in Dedham, Massachusetts, Papa Gino’s, Inc. is the parent company of the Papa Gino’s Pizzeria and D’Angelo Grilled Sandwiches restaurant chains. The company operates more than 275 company-owned and franchised Papa Gino’s, D’Angelo and dual-location restaurants and employs more than 5,000 people across New England. Both D’Angelo and Papa Gino’s are committed to providing high-quality products, attentive service, clean, convenient, attractive restaurants and a premium value experience for every guest.

To maintain competitive advantage and support business growth, the Papa Gino’s senior management team decided to rethink its information technology strategy and invest in innovation. The company’s IT team was tasked with assessing the existing technology landscape and identifying the areas where new systems and processes could deliver the greatest business value.

“At that time, we effectively had three main business systems: our JD Edwards ERP system, our restaurant point-of-sale system and Microsoft® Excel®, which we used very extensively,” explains Martha Lieber, Director of Business Systems at Papa Gino’s. “Basically, almost everything that wasn’t covered by the other two systems was managed in spreadsheets, which meant we had numerous sources of ‘truth’ across the organization. For company-wide processes like budgeting, this was a major problem, because it could take up to four months to consolidate and validate data from all the different spreadsheets and complete the process.”

Paul Valle, CIO, adds: “We knew that if we could create a single central solution for business analytics, we could standardize and significantly accelerate our planning and reporting processes. We also realized that it would allow us to analyze operational performance in areas that we had never previously been able to measure accurately – such as pizza

delivery times and marketing campaign effectiveness. When we presented our ideas to the board, they immediately agreed that this should be the top priority in our IT transformation.”

“We do a lot of email, SMS and postal marketing where we send coupons to our customers with special deals and offers,” explains Paul Valle. “With Cognos, we can track which coupons are actually being used, so we can see which kinds of deal are most attractive to customers and create the most sales. We recently created a loyalty program for regular customers which provides rewards that can be used on future orders and, thanks to Cognos, we have been able to track its effect on sales. What we’ve found is that customers who are members of the loyalty program order from us about 33 percent more frequently than those who haven’t joined yet. So this insight has led to a real drive to encourage customers to sign up.”


• Consolidate all the data

• Optimize labor costs while ensuring that restaurants have the right number of staff.

• Track the time it takes to deliver orders to customers’ homes compared to the estimated time

• Identify the effectiveness of the loyalty program

• Identify the order method that works best for the company so that this order method can even be promoted further


1. What were the problems of Papa Gino’s, Inc.? 2. What are the data available that can be used

for an analytics solution? 3. What are the possible solutions that can be

provided for Papa Gino’s, Inc. to help them solve their problems?

_________________________________________


May 2013


7

Case Analysis 4: Boerse StuttgartCase Analysis 4: Boerse StuttgartCase Analysis 4: Boerse StuttgartCase Analysis 4: Boerse Stuttgart


Founded in 1860, Boerse Stuttgart now employs 300 people and is Europe’s leading stock exchange organisation for private investors. With an average share of about 31 percent of the German market (March 2011), the company is Germany’s second largest trading centre. In 2009, its overall turnover reached €105 billion, exceeding last year’s result by about 23 percent.

With more than 684,000 listed securities, the exchange has achieved an average daily trading volume of about €503 million since the beginning of 2011 (March 2011).

As a publicly supervised stock exchange, Boerse Stuttgart AG is subject to numerous legal requirements. Compliance with regulations requires complete traceable documentation of all business processes. As a result of strong organic growth and several recent acquisitions, the company was using a variety of solutions whose data could not be easily analysed without a high level of manual effort. In particular, it was very difficult to perform ad-hoc-analyses based on current business data to support the market control department in making strategic and operational decisions.

Persistent business growth also meant that the data volumes that needed to be processed daily were perpetually increasing. Moreover, new internal and external requirements for reporting and analysis were evolving, and it was becoming difficult to fulfill these requirements cost-effectively with the available methods and tools.


• Consolidate all the data

• Comply to varying reporting requirements


1. What were the problems of Boerse Stuttgart? 2. What are the data available that can be used

for an analytics solution?

3. What kind of system can be built given the consolidated data?

4. How can business analytics culture be built in the company?

_________________________________________


May 2013


8

Case ACase ACase ACase Analysis 5: First Tennessee nalysis 5: First Tennessee nalysis 5: First Tennessee nalysis 5: First Tennessee BankBankBankBank


The company utilizes the First Tennessee Bank name in the state of Tennessee, its primary market and where it has dominant market share. Outside Tennessee (with the exception of suburban Memphis within Mississippi and Arkansas, and suburban Chattanooga in Georgia), it utilizes the First Horizon Bank nameplate, the company's official name. First Tennessee is the only major bank headquartered in the state. Until recently, it had expanded into Northern Virginia (2003), Maryland (2003).

The bank is a full-service provider of financial products and services for businesses and consumers.

On the path to becoming the Chief Marketing Officer of First Tennessee Bank, Dan Marks has shown all the earmarks of being a “numbers guy.” There’s the natural sense of skepticism—a need to see the proof in the numbers—that comes along with being the analytical sort. Do bank marketers have a unique way of thinking, of looking at the world? Marks affirms his belief that they do, but with a caveat. “Within banks, marketing people tend to be great at seeing opportunity and conceptualizing new ideas,” explains Marks. “But in today’s banking market, it’s increasingly important to view them with a critical eye, to discern where it makes most sense to focus the bank’s resources. That’s the balance—between creativity and discipline, between art and science—that we need to strike.”

What Marks had in mind went beyond budgets to the very heart of First Tennessee’s marketing processes, with granular metrics and analytics providing the basis for optimization. “Our aim was to shift from the ‘marketing-as-an expense’ mindset to the idea that marketing is a true profit driver.”

The market that Marks is talking about is defined not only by increasing competitive intensity, but also by the strategic challenges it presents to banks, not least of which is how to optimally focus marketing resources. Banks offer a more diverse portfolio of services than before, and they do so over a wider range of channels. While this trend has given banks more latitude to compete, it has made formulating and modulating marketing strategies, tactics and programs considerably more complex. That’s because as the range of options has grown—a good thing by any measure—marketing resources are as scarce as ever. To optimize how they invest them, banks need a way to continually measure their effectiveness, learn what works and adapt over time.

For banks today, having more ways to communicate with customers is a good thing. But it has also made it harder for banks to figure out where and how to most profitably commit their marketing resources. Leveraging predictive analytics, First Tennessee Bank is applying the ultimate acid test. It’s combining a granular understanding of the needs of customer segments with real P&L data to optimize its marketing spend, focusing on programs that deliver the highest ROI. First Tennessee’s ability to target its campaigns more intelligently has increased its response rate by 3.1%, cut key marketing costs by nearly 20%, and enables the bank to get the most from its resources.

Selling First Tennessee’s President of Banking on the idea wasn’t hard. Putting it into place required action on a number of fronts. Conceptually, think of an upside-down pyramid, with the technical capability to do predictive analytics at the bottom, underpinning the effort. Though it’s an essential foundation, its enablement represented a relatively modest share of the effort. More extensive (and the next level up in the pyramid) was the need to secure the business intelligence (from the bank’s data warehouse and finance organization) and from that develop the framework for ROI-based modeling. Based on product parameters like fees, spread and account balances, the model would create tiers of profitability for different kinds of products and accounts and—by extension—different segments of customers.

At the top of the inverted pyramid was Marks’s biggest challenge and, in many ways, the most important ingredient to his success— driving change at the process level. In this context, the key processes relate to choices made by marketing managers and within the lines of business around program funding.


9

Should, for example, more funding be put into customer acquisition activities like lead geration? Or should the funding emphasis be on retention and cross-selling opportunities? Traditionally, such decisions were the epitome of “in-the-box thinking”—guided by a mix of past experience, intuition and conventional assumptions about where the opportunity is.

Marks’s aim was to promote a new way of thinking about opportunity that would permeate the entire organization. “The message we’re getting across to our people is a new way of managing and optimizing our marketing resources around ROI—one that looks ‘under the covers’ at the relative profitability of all these programs and uses that as the basis for decision-making,” says Marks. “We understand that the most effective way to drive this change is not ‘top-down,’ but through a dialogue that moves all of us—including the LOBs—towards this new way of thinking.”

First Tennessee used to structure its marketing campaigns around product lines. Over the last few years, the bank has developed a highly systematic and targeted approach. Here’s how it works. Start with a granular understanding of each customer’s banking needs drawn directly from lots of customer data points. Using predictive analytics models developed by Marks’s staff, each customer is “scored” on their likelihood to purchase each product in First Tennessee’s portfolio. The corollary benefit of this approach is that it helps the bank’s marketers to pinpoint product clusters that represent “sweet spots” for cross-selling opportunities. That’s just the beginning. What sets the First Tennessee approach apart is how it applies a rigorous, systematic approach to prioritizing which opportunities make it to the campaign stage. By combining product revenue and cost information from its data warehouse with the segment data discussed above, First Tennessee’s model generates a quantitative measure of the expected profitability of a given product offered to a specific subset of its customers, such that each product/segment offer under consideration is assigned an expected ROI value. With this hierarchy established, Marks and his team now have an evidenced-based framework through which to prioritize programs and allocate resources accordingly.

The bank’s performance numbers also underscore the effectiveness of ROI-based optimization. For instance, the rate of response to its marketing campaigns has risen 3.1 percent, a reflection of its ability to more accurately target offerings to specific customer

segments based on their needs. First Tennessee’s recent growth in market share, producing gains across most of its geographic footprint in and around Tennessee, provides yet another window on its success. Overall, the bank has tallied a 600 percent return on its investment in predictive analytics through more efficiently deployed resources. By gaining the ability to target the most attractive segment for specific offers—a “qualityover- quantity” approach—First Tennessee has been able to optimize its campaign expenditures, evidenced by a 20 percent reduction in mailing costs and a 17 percent reduction in printing costs. Looking down the road, Marks expects tight resources and intense competition to be par for the course in the banking market. But with predictive analytical capabilities in place, he sees these conditions as only heightening the bank’s hunger for opportunity, since it has the ability to pursue it efficiently. “We’re committed to profitability and were committed to strengthening our customer relationships through every aspect of the way we do business,” says Marks. “Predictive analytics gives us the intelligence and insights we need to follow through on this commitment.”


• Identify areas where marketing resources should be allocated

• Identify which services to offer to clients


1. Identify how predictive analytics was used to solve the business problem. Explain how the predictive analytics solution works.

2. What were some of the predictors used for the predictive analytics model?

3. How did Dan Marks build an analytics culture?

_________________________________________

© Copyright IBM Corporation 2013 IBM Philippines IBM Plaza, Libis, Quezon City May 2013


10

Case Analysis 6: MoneyGram Case Analysis 6: MoneyGram Case Analysis 6: MoneyGram Case Analysis 6: MoneyGram InternationalInternationalInternationalInternational


The MoneyGram brand is recognized throughout the world as a leading global payment services company. The diverse array of products and services we offer enables consumer and businesses to make payments and transfer money around the world. From New York to Russia or London to India - in more than 197 countries - MoneyGram's money transfer service moves money quickly and easily around the world.

The payment services also help businesses operate more efficiently and cost effectively. We offer our products and services to consumers and businesses through a worldwide network of agents and financial institution customers.

Ted Bridenstine, systems development manager at MoneyGram—a leading global payment services company—underscores the importance of fraud detection with the story of a 100-year old grandmother who had contacted MoneyGram after receiving a call that her grandson had been arrested and needed US$2,500 for bail.

Behind the scenes, MoneyGram’s fraud detection system flagged the transaction as suspicious. Analysts determined that it was likely part of a telephone scam and a MoneyGram representative contacted the customer to let her know that the wire had been stopped and her money was being refunded. Worried about her grandson’s safety, she threatened to take her business elsewhere if MoneyGram didn’t wire the money. The company representative implored the woman to contact a family member and verify the story.

“She called back three days later in tears to thank the representative personally,” recalls Bridenstine. “She did verify that it was a fraud. She lives on Social

Security and could not afford to lose money. The call was emotional and heartwarming.”

Working with IBM and Alpine Consulting, MoneyGram has gained clear visibility into the transaction history of each customer and insight into “who’s who?”, “who knows whom?”, and “who does what?” as it analyzes money transfers. The company’s new fraud prevention system (known as the Global Compliance project during implementation) helps stop fraud in its tracks and reduces the overall time and work required to respond to new regulatory mandates, such as new requirements for International Automated Clearing House Transactions.

The system scans each transaction looking for signs of fraud and identifies suspicious or high-risk transactions based on established criteria. If fraud is detected, the system alerts analysts, who place the transaction on hold until a representative can confirm whether the transaction is legitimate or fraudulent. If fraud is detected, the company refunds the money to the sender.

Once accurate identity is established, the system can determine whether people are, or ever have been, related in any way and apply complex event processing to evaluate all transactions of the entity and of associated entities. The rules engine also provides the business logic to alert staff when a transaction or the aggregation of a specific customer’s transactions exceeds a specific threshold. Additionally, the solution gives MoneyGram the ability to respond quickly to new and different kinds of frauds. For instance, in 2010, MoneyGram noticed that fraudsters were receiving transactions in California. In early November 2010, MoneyGram analysts added a rule to the system that flagged transactions above a certain dollar amount sent to California. That simple change prevented US$1.7 million in suspected fraudulent transactions in just three months. “We can now react within hours or minutes, changing rules or implementing new rules, if an analyst identifies a new pattern of behavior,” says Bridenstine.

This new flexibility is helping the company increase customer satisfaction and stay a step ahead of fraudsters. Consumer complaints of fraud in January 2011 compared to January 2010 dropped 72 percent, with the most significant reductions in Canada, Nigeria, the United States, and the United Kingdom. From February 2010 to January 2011, the anti-fraud technology accounted for a 40 percent increase in


11

MoneyGram’s ability to identify and interrupt potentially fraudulent transactions.

“We are able to detect and respond to fraud faster than before to protect our consumers as well as our global network of agents,” says Bridenstine. “We must remain ever vigilant in the face of more sophisticated financial fraudsters. The solution stopped US$30,000 of fraudulent transactions on the first day and within 17 days of operation, it had stopped US$1 million of fraudulent transactions. We’ve already stopped more than US$37.7 million in all in fraudulent transactions and prevented thousands of customers from losing funds to fraud.”


• Detect signs of fraud and money laundering



2. What are the data used to make the predictive analytics solution work?

_________________________________________


May 2013


12

Case Analysis 7: HamiltonCase Analysis 7: HamiltonCase Analysis 7: HamiltonCase Analysis 7: Hamilton County County County County

Department of EducatiDepartment of EducatiDepartment of EducatiDepartment of Educationononon


The Hamilton County Department of Education is a diverse school system, providing roughly 42,000 students with a world class education. The system is nationally renowned for urban school success as well as middle school and high school reform. The district offers a variety of educational programs for all students including magnet schools, career academies and gender-based classes as well as a focus on individual student success and the goal of preparing all students to compete in the Global Economy.

It started a few years ago, when Hamilton County began looking into why its students were consistently scoring below state benchmarks on standardized tests. Administrators didn’t get far before realizing they lacked the kind of detailed, granular data that would be necessary to understand the factors that contributed to the poor performance, much less act on the problem. What little performance data Hamilton County had been receiving came from state scoring reports (as part of No Child Left Behind), which provided a lumpy, aggregated measure of whether the county’s 40,000 students were on track.

The other key indicator Hamilton County tracked on its own was the share of its students that graduated from high school. Kelly and his colleagues realized all too well that dropout rates and student performance were affected by many of the same factors—in some ways, two sides of the same coin. But most of all, Kelly’s experience—gained as a teacher and a principal earlier in his 18 years in the Hamilton County system—taught him that beneath the numbers were children, and that solving tough problems like

dropout rates and lagging performance required the county to address these problems on the individual student level.

But that meant first finding the students who needed help, and doing so before their problems led them to drop out of the Hamilton County school system. At the end of the day, there’s no one better positioned to sense when the student is in academic trouble than a teacher or counselor with whom students have a direct relationship. The trouble is, warning signs are often complex and cumulative in nature, thus escaping the notice of frontline educators and administrators. Add

to that the everyday challenges of running a classroom, and it’s all the more understandable how problem students can fly “under the radar” until it’s too late.

Kelly recognized that early detection of at-risk students required a more multidimensional view of their progress, performance and path through the Hamilton County school system. To accomplish this, Kelly developed a performance modeling tool that extracts individual student data from the county’s 78 schools and uses it to create predictive profiles, which help to flag those students in need of proactive intervention by teachers or counselors. Using built-in algorithms, the model determines which factors are the strongest predictors of a student failing or dropping out. Based on the outcomes of the model, each student is placed into one of four performance categories. Identifying those students labeled “fragile” or “off-track” is just the beginning of a process whose ultimate aim is the success of the student.

Hilary Smith is a key part of that process. Based at the Howard School of Academics and Technology in Chattanooga, some 15 miles away from the home office, Smith is the Department of Education’s Lead Counselor. Howard is an inner-city school, long seen as epitomizing the kinds of problems— such as high rates of dropping out and disciplinary problems—that Hamilton County is trying to address. When Smith first came to Howard seven years ago, it graduated just one in four students, a far lower graduation rate than the county as a whole.

For Hamilton County as a whole, that difference is already apparent in a graduation rate that has increased by more than 8 percentage points in the last year, to nearly 80 percent. Over the past few years, standardized testing scores have also increased by more than 10 percent for both math and reading. At the


13

Howard School, progress has been even more dramatic, with the graduation rate tripling over six years, culminating in a nearly 10 percent increase in the last year. “We went from one in four graduating to three in four graduating. You can’t ask for better than that in six years,” says Howard principal Paul Smith. “Though we’re not satisfied until we’re at 100 percent, we’re happy with the progress.”

While educational intelligence has largely been directed toward identifying and helping individual students, Hamilton County is also incorporating predictive analytics into day-to-day teaching and learning activities. Educators have always known that the roots of poor performance and drop-out behavior in high school often reach back far into a student’s early grades. Analytics is enabling Hamilton County educators to better understand how these adverse patterns form over the student’s academic life cycle—and what they can do to correct them. It’s seen in the way Hamilton County teachers have formed networks to more intensively interact with each other both across grades (e.g., high school and middle school) and within grades (e.g., a math teacher interacting with a social studies or English teacher) in the interests of individual students. A comprehensive and dynamic view of the student—crossing educational boundaries—is what makes this possible.

Hamilton County is also leveraging analytics to create innovative teacher incentive programs. By looking at the historical relationship between eighth-grade test scores and high school exams, the county is able to predict, in effect, a baseline performance benchmark for each student. By compensating teachers based on their ability to beat this rigorous benchmark, Hamilton County is using predictive analytics to encourage performance improvements for both teachers and students. Analytics is also providing Hamilton County administrators with the means to pinpoint opportunities to adjust the curriculum to meet the needs of a specific subset of the student population. For example, when analysis showed that male students were scoring below females on the state’s writing assessment test, Hamilton County responded by implementing a system-wide approach to address the needs of male students and close the performance gap.

Business ObjectivesBusiness ObjectivesBusiness ObjectivesBusiness Objectives

• Identify students who have a high risk of dropping out of school

• Identify special needs of subgroups of students

• Identify a benchmark to compare the performance of the students against to evaluate the performance of the teacher



2. Identify how analytics culture was built in the company.

3. Identify another possible predictive analytics solutions can be applied in schools. Identify the data that you need to be able to provide this predictive solution. Explain how this predictive analytics solution works.

_________________________________________


May 2013


14

Case Analysis 8: North Carolina Case Analysis 8: North Carolina Case Analysis 8: North Carolina Case Analysis 8: North Carolina State UniversityState UniversityState UniversityState University


The essence of a university is more than education—it is the advancement and dissemination of knowledge in all its forms. For leading research institutions like North Carolina State University, part of the university’s mission is fulfilled by the Office of Technology Transfer (OTT).

The function of the OTT is that of a sophisticated matchmaker. Its small staff of highly trained licensing professionals—many of whom have advanced degrees and are technology experts in their own right—seeks to find opportunities to license and ultimately commercialize NC State’s knowledge assets. These are usually promising new technologies or inventions created at the university, covering every field of endeavor from materials to biomedicine and more. “Our goal is to see actual consumer benefit come from the research conducted at NC State,” says Billy Houghteling, executive director of the OTT.

The six licensing professionals in the NC State OTT are very good at what they do. Today, over 100 products in the marketplace are based on work done at NC State.

Those successes represent the cream of the crop. Each year, more than 150 new developments come in to the OTT, to add to the university’s store of approximately 3,000 technology assets. Each is reviewed by one of the staff, and the most promising are pursued in depth. The professional assigned to the offering works to become a subject matter expert on the technology, then starts looking for companies that might be interested in licensing, developing and marketing it.

“We need to find the optimal partner for each case,” Houghteling says. “This can be difficult and challenging, because the market opportunities for any given asset are not always clear. It’s a complex triage process; we need to not only understand the technology itself, but also its commercial potential. And then we need to find the partner that can realize that potential effectively. It goes back to our purpose… it’s not just to license, but to see success in the marketplace.”

The process often takes from four to six months. Given the volume of new technologies being submitted to the OTT and the size of the staff, it can be virtually impossible for the office to keep pace. Added to this heavy workload is ongoing financial pressure brought about by a difficult economic climate. As a state institution, the university is facing austere times and the extra revenue brought in by technology licensing is of great benefit.

One of the main issues faced by the OTT licensing professionals is dealing with the vast number of data sources available to them. Online databases, websites, publications and much more have to be searched to find that proverbial needle in a haystack—the perfect development partner. “This can be a truly daunting challenge,” says Houghteling. “It’s just not practical for us to search every possible source of information.

”For example, a great deal of valuable information about the marketplace and potential development partners can be found in the myriad reports and databases provided by the Securities and Exchange Commission. Unfortunately, the potential return on the time it takes to find the right places to look is so low that it’s not practical to dig very deep.

The result is that good opportunities may go by the wayside. Ironically, the OTT professionals are fully capable and qualified to find them; they simply don’t have the time or resources to uncover those potential partners. “What we really need is analytics to help us,” Houghteling notes. “With the right tools, we can make that part of the process a great deal more efficient, which translates directly to more opportunities to see university research reach the marketplace.”

“We had a very clearly defined set of requirements,” Houghteling states. “Speed, of course, since that was a key challenge. But equally important was utility; speed is nothing if the answers are not useful to the licensing professional, so the output needed to be easily consumed and relevant; free of clutter and unrelated


15

leads. The system also had to be user-friendly, to reduce workload. And of course, it had to produce better results than we could achieve on our own.”


1. Identify what the business objective is. 2. Identify how predictive analytics was used to

solve the business problem. Explain how the predictive analytics solution works.

3. How can analytics culture be built in the company?

4. Identify another possible predictive analytics solution similar to this that can be helpful in another industry.

_________________________________________


May 2013


16

CaseCaseCaseCase Analysis 9: XO Analysis 9: XO Analysis 9: XO Analysis 9: XO CommunicationsCommunicationsCommunicationsCommunications


In the communications sector, customer loyalty is the key to profitability – it costs much less to retain an existing customer than it does to win a new one. As part of an organizational business transformation project, XO Communications identified an opportunity to improve its performance in the small- and medium-sized business sector by increasing customer retention, and needed a strategy that would help it achieve this in a cost-effective way.

The majority of the customers in this segment spend less than $1,000 per month with XO Communications, so it was not economically prudent to allocate dedicated client service managers to each account. At the same time, it was vital to offer these clients a high level of service and ensure that any issues they had were dealt with quickly and effectively. Instead of a ‘brute force’ approach – simply hiring more client service managers – the company needs to identify the customers who were most at risk of changing providers, and prompt the client service team to intervene appropriately.

The XO Communications team worked with IBM Education Services to build up a high level of competence in-house and transitioned away from the early models built by third-party consultants. This enabled the team to identify the factors that indicate whether a given customer is likely to change providers, and build a sophisticated statistical model that provides a monthly risk assessment for each customer.

Customers are prioritized based on the score the model calculates for them, and the client service team deals with the highest-risk customers first: making contact with them, finding out if they have any problems or are dissatisfied with the service they are receiving, and then taking appropriate steps to improve the situation.

By providing an analytics solution in-house and giving client managers the ability to identify the customer relationships that need most attention, XO Communications has been able to reduce customer churn without substantially increasing headcount.

XO Communications has applied the skills and experience from this project to generate additional returns in subsequent projects. With additional tuning and analysis, the next phase of the customer churn project has produced another $7 million in revenue and the model is 50 percent more accurate than the original. Not only has this prevented churn, but it has also increased the revenue stream. Other areas of focus include profitability analysis, bad debt write-offs, and some further levels of granularity with predictive modeling on a per-product basis.

The project originated as part of a full-scale transformation initiative to increase business efficiency at XO Communications. During the consultation phase of this initiative, the company decided to look into an analytics solution, and began a small project with a third-party consulting partner. The results of this project convinced XO Communications that analytics could be a vital tool, and also highlighted the benefits of building up a capability in-house.

The company formed a Customer Intelligence team, tasked with purchasing and building a solution. The team performed a very thorough vendor evaluation before selecting IBM SPSS Modeler and IBM SPSS Statistics. This decision was based on the perceived superiority of the SPSS products in terms of functionality, scalability, licensing terms and ease of use.

The team utilized the Cross Industry Standard Process for Data Mining (CRISP-DM) methodology to manage the project through six main phases: business understanding, data understanding, data preparation, modeling, evaluation and deployment. During the process, the team evaluated more than 500 variables to see which had the most influence on customer retention. Ultimately, a model was built based on the 25 most relevant variables. The model assigns risk scores to each customer. The top 20 percent reflects 66 percent of the churn, so the top 10 percent are passed on to the centralized retention team, and the next 10 percent are passed to the field retention team for remedial action.

As part of the project, the Customer Intelligence team received comprehensive training from IBM, which helped not only to gain technical knowledge, but also to build up a set of best practices for future predictive analytics deployments. As a result of the expertise gained, the team’s churn prediction model has been able to outperform the early third-party model by 50


17

percent. With the success of this initial project, the team is now embarking on a number of new initiatives using predictive analytics, as well as expanding information delivery with dashboards and business intelligence.


1. What analytics solution is needed by the company? What are the necessary components to solve the problem?

2. Why should analytics be used instead of just hiring more client service managers?

3. Given the CRISP-DM approach that makes use of the following steps in providing a predictive analytics solution:

a. Business Understanding b. Data Understanding c. Data Preparation d. Modeling e. Evaluation f. Deployment

Give the phases that were described in the case study given.

4. Identify how the opportunity was identified and explain why this solution was feasible.

5. Explain why data understanding and data preparation are important in this analytics solution.

_________________________________________


May 2013


18

Case Analysis 10: Clark CountyCase Analysis 10: Clark CountyCase Analysis 10: Clark CountyCase Analysis 10: Clark County Family Services Family Services Family Services Family Services

CompanCompanCompanCompany Backgroundy Backgroundy Backgroundy Background

In 2004, Clark County Department of Family Services was given responsibility for all child welfare, foster care and adoption services in the county. As part of a federal improvement plan for the State of Nevada, the department also needed to be able to comply with rigorous reporting requirements. To gain a true overview of performance and manage workload effectively, it would be necessary to unite data from numerous systems into a single source. With a growing population and an increasing case-load, the demand for easily accessible data became even more acute.

It was critical to find a solution that could extend the investment in the ERP and GIS applications, could scale quickly to support all functional areas, and included a broad range of capabilities for metrics, dashboards, reports, analysis, and self-service information access. A crucial element in this initiative would be to make the solution easy for non-technical users to work with – moving workload away from the central IT team and enabling different business areas to serve themselves with information.

After a rigorous selection process, Clark County decided to implement a business analytics solution that would provide a single source for all family services-related data. The purpose was to enable users to gain deeper insight into key operational areas such as intake, placement, investigations, permits, adoption, court visits, eligibility for funding, payments, recruitment and training.

Users within each of these operational areas can now answer the key questions that support their day-to-day decisions. For example, members of the intake team can instantly view key metrics such as how many referrals have been made on the child services hotline, how many of them need to be investigated, and how many can be ruled out. Almost 100 percent of these analyses are now handled by the departmental users themselves, with no need to ask for help from the IT team.

Clark County implemented a phased deployment of business analytics technologies to meet both immediate business needs and long-term goals. The

initiative started with the creation of a project team that represented user across the whole organization, at every level from report-writers to senior executives. This team set the key objectives – implement the business analytics solution to deliver the required capabilities for metrics, dashboards, reporting, analysis and self-service information delivery. The solution would extend the investments in the SAP ERP and GIS applications, support and data warehouse architecture, and be architected for rapid, enterprise-wide deployments. PerformanceG2 was chosen to translate the wide range of business requirements into the business analytics solution.

The first phase of the implementation involved business-led prioritization. Creating scorecards and metrics with underlying OLAP cubes for the intake, placement and investigation teams were among the first objectives. These helped managers not only monitor performance more effectively but also drill down to investigate underlying root causes. Phase two built on the experience of the first phase. A change in emphasis at the senior management level led to an increased focus on creating a more comprehensive data warehouse. At phase completion, information access was extended to every functional area across the department. Business users had self-service access to information, and were able to develop reports for themselves without IT assistance. The development of the data warehouse supports data demands. Today, the culture has shifted and nearly 100 percent of the department’s reports are created and managed entirely by business users.

Questions Questions Questions Questions

1. What type of analytics solution is needed? Identify the opportunity for an analytics solution and explain why this solution is feasible.

2. Identify the source of data for the analytics solution needed.

3. Who are the business users? 4. What problems should be addressed by the

solution? 5. Explain the analytics solution implementation

flow for this case.


19

Case AnalysisCase AnalysisCase AnalysisCase Analysis 11: Memph 11: Memph 11: Memph 11: Memphis Police is Police is Police is Police DepartmentDepartmentDepartmentDepartment


With traditional policing practices unable to thwart a rising rate of criminal activity and budgets tight, the Memphis PD pioneered a way to focus their patrol resources more intelligently. By recognizing crime trends as they are happening, MPD’s predictive enforcement tool gives precinct commanders the ability to change their tactics and redirect their patrol resources in a way that both thwarts crimes before they happen and catches more criminals in the act. Through such smart policing approaches, MPD has reduced the overall crime volume in Memphis by 30%, making life safer for the citizens that were demanding it.

It all started when Larry Godwin, Director of Memphis Police Services convened what would become a landmark meeting of Memphis’s law-enforcement A-Team, with the purpose of stimulating fresh ideas on how to reverse a rising tide of crime. In what might be called the “cafeteria summit,” Godwin and key members of the department’s Organized Crime Unit (OCU) sat down with District Attorney General Bill Gibbons (whose district included Memphis) and Dr. Richard Janikowski, a professor of Criminology at the University of Memphis. Over a sensibly priced meal served on trays, Godwin sketched out a scenario of rising crime, frozen (or even shrinking) budgets and a growing disenchantment among Memphis citizens—and was open to ideas. As Director of the university’s Center for Community Criminology and Research, ideas were Janikowski’s specialty. Over the decade before, he had been involved in a number of analytical initiatives into better understanding crime patterns. Now, with the MPD requesting his input, Janikowski saw the opportunity to put into practice the simple yet powerful principle that “If you focus police resources intelligently by putting them in the right place, on the right day, at the right time—good things are going to happen,” says Janikowski. “You’ll either deter criminal activity or you’re going to catch people.”

Godwin liked what he heard. So much so that he agreed to regularly share key crime data with Janikowski and his colleagues—a gesture that goes against the deeply ingrained tendency for police departments to hold their information close. Using this crime data, Janikowski’s job was to develop an analytical framework that would be used as the basis for a pilot program, the results of which would shed light on which analytical and operational approaches worked and which didn’t.

A few months later, that effort materialized into a three-day operation that proved to be one of the most effective ever. By identifying hot spots at a granular level, MPD made some 70 arrests in just the first two hours—a number usually made on an average weekend—and went on to make a total of 1,200, with crimes ranging from drugs to weapons charges to prostitution and other “quality-of-life” offenses. It was a great start, but only a start. Godwin realized that over the long-term, the success of the program would require not only predictive analytics capability but also the adaptation of the department’s operational processes to take full advantage of them. Godwin further realized that moving from a pilot project to a systemic change in practices would require broad buy-in, especially from patrol officers out on the street. It’s not only a question of communicating how predictive modeling can help our officers be more effective, says Godwin, but also knowing how to listen to them and tap into their knowledge. “Nobody knows a ward better than the patrolman who rides as many as six or seven days a week for eight to 10 hours a day,” says Godwin.

“Showing our willingness to learn from their knowledge and experience is the best way to get them to take ownership.” To secure mayoral approval to move ahead with the program, Godwin prepared a business case that resonated with the brutal budget realities that Memphis shares with most major American cities—the need to confront a growing problem with fixed or shrinking resources. It was widely acknowledged that the MPD needed to add another 500 patrol officers to offset a growth in criminal activity, but that would take nearly 6 years to achieve. Godwin’s aim was to show how the intelligent alignment of police resources would effectively enable the department to close the manpower gap now—a must in the eyes of Memphis’s citizens. Under the plan Godwin proposed, each precinct commander in the MPD would be given the resources (in the form of overtime funding) and


20

flexibility to make their own deployment decisions based on intelligence provided by the solution. Most importantly, results would be rigorously measured and commanders held accountable for their performance. It didn’t take much selling, because a few hours later, Godwin and the mayor were standing in front of the press touting the newly approved program—which came to be known as Blue CRUSH—as a way to intelligently reduce crime.

Muscular connotations aside, Blue CRUSH (Criminal Reduction Utilizing Statistical History) is really about gaining advantage through insight and agility. At the heart of it is a predictive model that incorporates fresh crime data from sources that range from the MPD’s records management system to video cameras monitoring events on the street. In the realm of crime-fighting analytics, there’s a fine line between the “interesting” and the actionable. It is strength in the latter that makes Blue CRUSH stand out from its predecessors. Blue CRUSH lays bare underlying crime trends in the way that promotes an effective fast response, as well as a deeper understanding of the longer-term factors (like abandoned housing) that affect crime trends. It happens at the precinct level. Looking at multilayer maps that show crime hot spots, commanders can see not only current activity levels, but also any shifts in such activities that may have resulted from previous changes in policing deployment and tactics. At each weekly meeting, commanders go over these results with their officers to judge what worked, what didn’t and how to adjust tactics in the coming week. They might see, for example, how burglaries are down in one ward, but up another, or where thieves are stealing cars in one ward and dumping them in another. What’s striking, says Godwin, is the granularity. “We’re catching this immediately and we’re doing it every day,” he explains. “On short notice, we’re able to shift officers to a particular ward, on a particular day, right down to the shift level. It’s a bit like a chess match and it’s enabling us to make arrests we never could have before.”

If there’s an unsung hero in the MPD’s success story, it’s accountability. The experiences of other departments in analytical police work—as well as the MPD’s early efforts—had shown Godwin the importance of rigorous and consistent reporting practices, employing common metrics, across precincts. Godwin conveyed this message to the department in two ways. The first was his decision to employ a standardized reporting template for all

commanders, thus discouraging the tendency to “cherry pick” results and obscure meaningful comparisons. Further reinforcing the message (and removing all ambiguity) was Godwin’s decision to rename the weekly sessions TRAC (Tracking for Responsibility, Accountability and Credibility) meetings. The fact that TRAC meetings are also a forum for precinct commanders to share their ideas—and, in many cases, learn from each other’s mistakes—is an outgrowth of the more open culture Godwin has tried to engender. The results of Memphis’s intelligent policing strategy speak loudly. Since Blue CRUSH was rolled out citywide, it has produced a sharp and sustained impact on crime rates in Memphis, including a more than 30 percent reduction in serious crime and a 15 percent reduction in violent crime. One recent enforcement action—targeted to drug dealers in a specific Memphis neighborhood—produced results reminiscent in scale of Blue CRUSH’s very first pilot operation, producing 50 arrests and leading to a 36.8 percent reduction in crime in the targeted area. In the MPD’s Felony Assault Unit (FAU), the department leveraged insights from Blue CRUSH to optimize which types of cases its officers needed to focus on. As a result of the subsequent realignment police resources, the FAU’s conviction rate rose fourfold, from 16 percent to nearly 70 percent.


1. Identify how the opportunity was identified and explain why this solution was feasible.

2. Explain why data understanding and data preparation are important in this predictive analytics solution.

3. What was used as basis for the pilot program? 4. What are the limitations of the predictive

analytics solution that was provided? 5. How was analytics culture built at Memphis

PD? 6. Where were the data used for predictive

modeling pulled from? 7. What are the predictors? What is the target? 8. What is expected from the predictive model?

_________________________________________


May 2013


21

________________________________________________________________________________

Licensed Materials - Property of IBM © Copyright IBM Corporation 2013 Published May 2013

IBM, the IBM logo and ibm.com are trademarks of International Business Machines Corp., registered in many jurisdictions worldwide. Other product and service names might be trademarks of IBM or other companies.

This case study set contains proprietary information which is protected by copyright. No part of this document may be modified without a legal license agreement from IBM Corporation. Any references in this information to non-IBM Web sites are provided for convenience only and do not in any manner serve as an endorsement of those Web sites. The materials at those Web sites are not part of the materials for this IBM product and use of those Web sites is at your own risk.

Documents

IBM Business Analytics Case Studies