14. Building a Global Analytical Capability
Thomas H. Davenport
Analytics are being used successfully by companies around the globe to reduce risk, uncover new growth opportunities, and make existing business lines more efficient and profitable. Although the use of analytics is expanding, in most multinational companies there is little coordination of the organization’s analytical initiatives. Substantial geographic variation in analytical approaches occurs within the same company. Some impressive analytical work takes place at corporate headquarters, and little goes on outside the home country—and sometimes vice versa. Teams typically operate independently within countries, business units, and functional areas, based on local conditions and requirements.
Based on conversations with analytics leaders around the world, few organizations are managing analytics globally. One reason is that not many organizations have anyone in charge of analytics at the global level. Rarely does a “Chief Analytics Officer” oversee analytical activity at a global level across all groups. More common is for different business units, functions, or country-based organizations to have their own analytics capabilities. Sometimes even within a country, siloed analytical groups can exist within functions or business units.
Widespread Geographic Variation
Even strong analytical competitors can have substantial geographic variation. Some examples may be useful. Take Tesco, for example. Based in the U.K., it’s the world’s third-largest retailer and has operations in 13 other countries. With the help of consultants dunnhumby (in which Tesco eventually bought a majority ownership share), the company pioneered the use of its loyalty card (ClubCard) data to target promotions to members. It’s been a fantastically successful program. It is responsible in large part for Tesco’s doubling its U.K. market share since Clubcard was introduced in 1995.1
However, there seems to be wide variation across countries in whether ClubCard—or an equivalent program—is offered and the extent to which its data is used to target promotions. It’s definitely not offered in the company’s U.S. Fresh & Easy chain, where loyalty programs are even somewhat disparaged on the website.2 Korea—where Tesco operates a number of superstores (initially in a joint venture with Samsung, but now wholly owned) under the Homeplus brand—uses both a loyalty card (Familycard) and the resulting data.
Another example of geographic variation is Banco Santander, the Spain-based bank that is now the world’s eighth largest in terms of assets. In Spain, there is a substantial focus on analytics, although the bank is not the market leader in that regard. In Brazil, Santander does not have a major focus on credit card analytics. Mexico, on the other hand, is quite aggressive on credit card analytics, basically emulating the very successful (in the U.S., at least) approaches of Capital One. In Germany, Santander has made major strides on credit scoring and automated loan decision models. However, at Sovereign Bank, the U.S. bank that Santander owns, there is little focus on analytics. The only global approach to analytics involves risk management, a consistent approach to which is somewhat mandated by Basel II regulations.
Is this geographic variation good or bad? One could argue that it’s somewhat necessary given that regulations and available information vary across the world. In Brazil, for example, there is no such thing as a credit score (only binary indications of whether a particular customer is worthy of credit), which limits the ability to make loans on that basis. A head of analytics at a bank commented that the inability to get the same types of data globally was the greatest impediment to a common global analytics strategy.
But there is an opportunity to do more to standardize analytical approaches. Tesco, for example, is beginning to try to create more consistency and aggressive use of analytics through its dunnhumby subsidiary. The company has appointed an “Analytical Ambassador” to spread analytics through global subsidiaries.
Global Coordination of Analytics
Tesco’s ambassador, however, is only the beginning of what can be undertaken in the realm of global coordination. Some other leading firms have taken a much more closely managed approach to coordinating analytical approaches around the globe. There are at least three possible global structure options:
• Central coordination through a common global analytics organization
• A strong center of excellence model that doesn’t own all analytical groups but attempts to coordinate their efforts
• A division of labor model in which different groups around the globe specialize in particular analytical approaches
The following sections describe each of these global coordination models and provide examples.
Central Coordination, Centralized Organization
The most aggressive approach to global coordination of analytical activity is a central corporate organization to support analytics. Like all centralized structures, it allows for efficient, common approaches to solving business problems. Analysts in remote areas far from headquarters can still have been recruited, trained, and supervised from a central analytics organization—even if they also have local reporting relationships.
Of course, central organizations also have a downside. Analytical problems that can’t be solved through common approaches may not get much focus. They may go unsolved or may require help from external consultants. The emphasis is on repeatable, standard solutions that can be solved through consistent models and software.
Procter & Gamble provides a good illustration of a centralized global coordination model. Several years ago, the corporate IT organization began to build and consolidate analytical people and renamed itself Information and Decision Solutions (IDS). Analytics were made a part of the Business Intelligence organization within IDS. A central group of analysts (divided into Commercial and Product Supply subgroups) at headquarters can work on a variety of solutions to be applied throughout P&G.
There are also analysts from IDS who support particular brands and geographic business units. The Asia business, for example, has someone responsible for business intelligence and analytics. The local analysts typically have close relationships with the leaders of individual business units and work on problems that the leaders have prioritized.
Using this central approach, P&G has rolled out a new approach to managing the business globally. It consists of more than 50 Business Spheres—special rooms designed for the display and discussion of business performance information and analyses. The information and analyses displayed are contained in a series of Business Sufficiency Models defined at headquarters, which continually evolve with new analyses and business challenges. It is unlikely that P&G could have developed and rolled out such a global approach without the global organization that supports it. Of course, the success of such an initiative also requires strong support from senior management, and CEO Bob McDonald has been a strong partner of IDS in the initiative.
A Strong Center of Excellence
A somewhat less centralized, but still effective, organizational model for global coordination requires the creation of a strong center of excellence to build and coordinate analytical activity. The actual center typically is composed of only a few people, and not all analysts report to them. But they have been deputized to take steps to build and acquire people with the necessary analytical skills, develop or subsidize the development of analytical solutions, and create structures for capturing and sharing analytical knowledge. Typically a head of analytics is at the top of the center, and there may be regional or unit-specific analytics heads as well.
This is a less centrally structured organization than the first one described, but it does require a substantial commitment to analytics. A firm couldn’t establish all the necessary structures and positions in a center of excellence if it didn’t believe that analytics were critical to its success. But compared to the fully centralized model, it does allow for greater focus on the idiosyncratic analytical problems of business units, geographies, and business functions that may be far from headquarters priorities. It also provides a structure for sharing analytical solutions and assets.
Perhaps the best examples of this approach come from professional services firms, several of which have established global initiatives and centers of excellence. Both Deloitte and Accenture, for example, have established global organizations to drive and support the development of analytical capabilities and the successful implementation of client projects. Deloitte, for example, has a global head of analytics (based in Singapore, but traveling widely) with a small staff. There are also heads of analytics in the Americas, Europe/Middle East/Africa, and Asia. In addition to these geographic organizations, there are also analytics groups with heads for particular practices (Audit, Tax, Consulting, Financial Advisory) and analytical domains (human capital, customer relationships, pricing, supply chain).
Although each group has considerable autonomy, annual global meetings are held to share ideas and solutions and report progress. An individual analytics practitioner may look to a variety of groups—only one of which involves analytics—for career direction and performance monitoring. However, this type of organization has allowed both Deloitte and Accenture—and several other firms as well—to quickly build analytical capabilities to serve external client needs.
A Coordinated “Division of Labor” Approach
Both of the organizational structures for global coordination just discussed require a high level of commitment to analytics and the devotion of considerable resources to building the organization. However, companies that don’t have as much management commitment can employ a third approach that involves some coordination but less formal structure. It is a largely decentralized approach that also involves a small degree of coordination and collaboration through a recognized division of analytical expertise and labor. Analysts primarily work on local problems but devote a fraction of their time to developing solutions that can be spread across the company.
For example, one global insurance leader has developed a unique model over the past five years in its non-U.S. business that has leveraged local successes across the entire organization. Decentralized analysts in “data labs” across the world spend the bulk of their time focused on applying analytics to drive the business in their local market. But these analysts spend 10% to 15% of their time identifying best practices that could be leveraged on a global basis. A global center of excellence (consisting of only a global analytics leader and one other analyst) based in Hong Kong compiles the learnings from these disparate data labs, packages them, and disseminates them. In this way, this insurance provider has locally focused analysts but has developed an integrated global analytics capability. This effort has proven to senior leadership in the global business not only that analytics produces a positive ROI, but also that it is now a durable, global competitive advantage for the company.
The global business for this insurer is a $2 billion business that sells four insurance product lines in 27 countries. This business often works with affiliate partners and uses about 6,500 telemarketers to sell its products globally. Its U.S. business, where headquarters is located, focuses primarily on one line of insurance products and has a centralized analytics group of almost 50 people. However, the U.S. analytics function has no responsibilities outside that country and coordinates only with the global analytics group informally.
The company began competing on analytics in 2005. Initial efforts focused on using data to more effectively manage customer relationships. Soon, the focus evolved much more toward increasing overall customer value. As the effort matured, customer value management (CVM) was chosen as the name of the non-U.S. group. Although subtle, this choice of name connoted in the minds of senior management a business function rather than a simple application of technology. The formal definition of CVM is “the art and science of measuring, analyzing, and managing customer value.” For this company, CVM is about testing, measuring, managing, and analyzing data to increase customer value (for both the company and its partners) while driving increased satisfaction for the end customer. In 2007, the insurance provider began working with affiliate partners in using intelligence generated from CVM to help these partners increase sales and commissions—work that continued through 2008 and 2009. The use of CVM has now extended across all of the company’s product lines. The company has used CVM to help win new business and as part of new-product launches.
A CVM Center of Excellence (CoE) was established in Hong Kong in 2010. This small centralized organization has the resources and capacity to define and disseminate global CVM best practices. At the same time, this insurance provider has established “country data labs” in countries across the world. On average these labs have about three analysts who spend about 85% to 90% of their time focused on using analytics and CVM in their local market. They spend the remaining 10% to 15% of their time focused on establishing global best practices that can be used by other analysts at the company in other countries outside the U.S. (or theoretically in the U.S. too).
In addition to their day-to-day responsibilities working with local affinity partners, the labs in each market have a particular area of analytical concentration. For example, the lab in Taiwan concentrates on randomized testing and learning. In China the data lab is focused on product optimization. In Spain, which is the company’s first market for a new private medical insurance (PMI) product, the data lab focuses on analyzing PMI. The learnings from each data lab are shared with the CoE in Hong Kong, which distills the findings and packages the key insights with the labs around the globe.
Now, five years into this journey, the CVM team can show positive returns. This insurer’s global capabilities provide a competitive advantage against strong local competitors in each market. These global analytical capabilities are attractive to the company’s partners because the capabilities provide insights that help improve commissions. The company’s partners have experienced positive returns from the company’s CVM capabilities, and they view CVM as a major differentiator. The use of analytics to build models results in better targeting of sales and marketing activities. Management strongly supports the CVM team, and the CVM analysts have unfettered access to whatever data they need to do their jobs.
Other Global Analytics Trends
In addition to these organizational structures for global coordination, I’ve abstracted a number of observations from travels and discussions with analysts and senior managers around the world:
• Analytics lags ERP implementation. In general, about five years after a firm implements an ERP system, it realizes it has a great deal of unused data, and some executive asks, “Weren’t we supposed to do something with this data to better manage the business?” This leads to increased interest in analytics to leverage the data. One executive from SAS noted that this trend applies at the country level as well. About five years after the country’s large organizations have completed ERP implementation, the business of analytics companies such as SAS picks up dramatically.
• Financial services companies are the most active users of analytics. Although there are analytical competitors in virtually all industries, the use of analytics in financial services is probably greatest around the world. In the wake of the financial crisis, it seems that financial companies are using analytics to help identify and manage their risks. Telecom, health care, and retail are other industries where analytics is beginning to take off around the globe. Governments are also developing a much stronger interest in analytics.
• Analytics is being used to reduce risk and pursue new opportunities with customers. Whether the primary use is risk reduction or pursuit of customer opportunity depends on the industry and the geography. In general, in Europe analytics are being used to mitigate risk and prevent customer attrition. In fast-growing economies in Asia, particularly in China, analytics are being used to capitalize on opportunities, and there is somewhat less interest in risk-oriented applications.
• Data privacy is a growing issue everywhere. Analysts and executives are concerned about data privacy and security in every region and country. They feel there is considerable uncertainty about how to deal with existing regulations, which vary widely across countries and even states within some countries. This uncertainty could affect the data that is available to analysts and how this data can be used. In addition, data privacy laws can change quickly, requiring that analysts and executives closely follow and stay abreast of these laws.
• American advantages. The United States has the most analytical competitors, perhaps because far more data is available in the U.S. for analysts to use. However, the lack of data in other geographies leads analysts to be creative and resourceful. The move to big data is also more pronounced in the U.S. than in other countries, although some Asian countries, including Singapore and South Korea, are displaying considerable interest in big-data analytics.
• Recruiting analysts globally. Other than in centrally managed analytics organizations such as Procter & Gamble (which regularly moves analysts through positions around the world), most labor markets for quantitative analysts are local. But finding people who can use data to tell compelling stories is quite difficult in every market. The approach at the insurance company described earlier has been to hire people with good analytical skills and then teach them about the business and how to use data to make a difference in the business.
• Country-based investments in analytics and big data. Over the last several years, we’ve begun to see countries that view analytics as key to their future economic development. Ireland, Singapore, and (very recently) the United States are examples of this trend. Analytics make sense as a growth area for Singapore for several reasons. Some of the country’s previous growth domains, such as information technology manufacturing, have become somewhat commoditized. Singapore has scored high (first or third) in every TIMSS (Trends in International Mathematics and Science Study) ranking of mathematics achievement since 1995. Many Singaporeans speak excellent English—the language most often used to discuss analytics in business. Singapore has a strong industry foundation in financial services, one of the most analytical industries. Finally, the country’s citizens are early and aggressive adopters of consumer technologies, which generate a lot of data for analysis. The government of Singapore has provided support for both university and private-sector programs. The Living Analytics Research Centre3 at Singapore Management University “seeks to make Singapore one of the world’s premier locations for the development and applied use of real-time consumer and social analytics.” The government also has funded other universities in areas involving big-data analytics. Furthermore, it supported the creation of the Deloitte Analytics Research Center in Singapore. In short, Singapore has decided that analytics are of sufficient promise as a driver of the nation’s future economic growth that it is subsidizing a substantial research program on the topic. The Irish and U.S. governments have announced similar programs of research and education support, but on a smaller scale than Singapore.
Overall, we are still in the early days of global coordination and management of analytics programs. We are certain to see new approaches, and new examples of the models described in this chapter, as firms become more advanced and sophisticated in their use of analytics. The trend, however, is toward more coordination and collaboration over time, rather than less.
Endnotes
1. For more on the operation and history of Clubcard, see http://en.wikipedia.org/wiki/Tesco_Clubcard.