Assessing the Evidence

Many researchers have found that fact-based decision processes are critical to high performance. In Good to Great, for example, Jim Collins notes that “breakthrough results come about by a series of good decisions, diligently executed and accumulated on top of another . . . [Good-to-great companies] made many more good decisions than bad ones, and they made many more good decisions than comparison companies . . . They infused the entire process with the brutal facts of reality . . . You absolutely cannot make a series of good decisions without first confronting the brutal facts.”⁷

Researchers have also begun to document the returns that companies can earn from investments in specific analytical technologies or initiatives. For example, technology research firm International Data Corporation (IDC) found in one study that analytical projects aimed at improving production had a median ROI of 277 percent; those involving financial management had a median ROI of 139 percent; and those focused on customer relationship management, a median ROI of 55 percent.⁸ The study also showed that the median ROI for analytics projects using predictive technologies was 145 percent, compared with a median ROI of 89 percent for projects without them.⁹ Nucleus Research, a company that researches IT value, reported in 2014 on a multiyear study of the return on analytics investments. It found that a dollar spent on analytics returned $13 on average.¹⁰

Similarly, the consulting firm Bain & Company surveyed four hundred large companies around the world in 2013, and asked executive respondents about their firms’ analytics capabilities. The consultants concluded that only 4 percent of the companies were “really good at analytics.” But those analytical competitors were:

• Twice as likely to be in the top quartile of financial performance within their industries
• Three times more likely to execute decisions as intended
• Five times more likely to make decisions faster.¹¹

To fill in the gaps of evidence about the effect of analytics on business performance, we conducted two surveys—the first an in-depth sample of thirty-two organizations that we rated in terms of their analytical orientations, and the second a much larger survey of firms that had made major investments in enterprise systems. In the first survey, we rated each firm’s stage of analytical maturity (the same five-point scale described in chapter 2, with 1 equaling major challenges to analytical competition and 5 indicating analytical mastery). Then we gathered financial performance data on all the survey participants. After conducting a statistical analysis of the data, we found a significant correlation between higher levels of analytical maturity and robust five-year compound annual growth rates.¹²

In the second study, we surveyed more than 450 executives in 371 large and medium-sized companies. We limited this study to those companies that had already implemented at least two modules of an enterprise system and therefore had a sufficient quantity and quality of transaction data available for analysis.¹³ Those companies represented eighteen industries in thirty-four countries.¹⁴ This study was a follow-up to an earlier study on the value of enterprise systems and analytics.¹⁵

In the large survey, we found a direct relationship between enterprise systems and decision making. For example, while many organizations initially invest in enterprise systems to improve efficiency and streamline processes, we found that cost savings were not their primary objective. In both studies, a majority (53 percent) of respondents identified “improved decision making” as one of their top three business objectives. To help managers make more-informed and faster decisions, organizations initially invest in enterprise systems to deliver reliable transaction-level data and “a single version of the truth”—an important precursor to developing an analytical capability. Once a firm has established a solid foundation of high-quality transaction data, its managers are able to shift their focus to using the data and systems for better decision making.

We also found that companies are becoming more analytical over time and building their commitment to analytics. In the first study, nearly half (45 percent) of the companies we surveyed reported that they had minimal or no analytical capabilities. However, four years later, only 8 percent said they lacked basic analytical capabilities. Almost every large firm has those capabilities today.

Similarly, the number of organizations with significant or advanced analytical capabilities supported by extensive and integrated management information doubled from 28 percent to 57 percent.

Most important, we found (and subsequent studies by other researchers has confirmed) a striking relationship between the use of analytics and business performance. When we compared the responses of high performers (those who outperformed their industry in terms of profit, shareholder return, and revenue growth—about 13 percent of the sample) with those of low performers (16 percent of the sample), we found that the majority of high-performing businesses strategically apply analytics in their daily operations. And about 10 percent of executives cited analytics as a key element of their company’s strategy. High performers were 50 percent more likely to use analytics strategically compared with the overall sample and five times as likely as low performers.

Further, we discovered a significant statistical association between an organization’s commitment to analytics and high performance. Companies with strong analytical orientations (those who answered with a 4 or 5 on all our questions) represented 25 percent of the sample (ninety-three companies), and their orientations correlated highly with financial outperformance in terms of profit, revenue, and shareholder return.¹⁶ In fact, one of the strongest and most consistent differences between low- and high-performance businesses is their attitude toward, and applications of, analytics (see figure 3-1).¹⁷ For example, 65 percent of high performers indicated they have significant decision-support or real-time analytical capabilities versus 23 percent of low performers. Only 8 percent of low performers valued analytical insights to a very large extent compared with 36 percent of top performers. And while one-third of the low performers in our study believe that they have above-average analytical capability within their industries, 77 percent of top performers believe that. Finally, 40 percent of high performers employ analytics broadly across their entire organization, but only 23 percent of low performers do.

Subsequent studies reached similar conclusions. In 2010, MIT Sloan Management Review and IBM analyzed data from three thousand executives in over one hundred countries and thirty industries. They concluded that “top-performing organizations use analytics five times more than lower performers. Top performers approach business operations differently from their peers. Specifically, they put analytics to use in the widest possible range of decisions, large and small. They were twice as likely to use analytics to guide future strategies, and twice as likely to use insights to guide day-to-day operations. They make decisions based on rigorous analysis at more than double the rate of lower performers.”¹⁸

After concluding a seven-year study of data from 864 respondents in nine countries and eight industries, Accenture and Professor David Simchi-Levi at the Massachusetts Institute of Technology (MIT), in 2015 concluded that “the stronger a company’s commitment to analytics, the higher that company’s performance” (see figure 3-2).

Among their findings are that, compared to low performers:

• Twice as many high performers are using analytics in key areas to support decision making.
• Twice as many high performers are embedding analytics in decision making that leverages machine learning.
• High performers embed predictive analytics insights into key business processes at higher rates and keep monitoring decisions and course-correcting as needed.
• High performers are three times as likely to invest a substantial portion of their technology spend on analytics. In addition, more than twice as many high performers are spending more on investments in analytical human capital through training, investments in people and the use of consultants.¹⁹

The breadth and consistency of the associations described earlier suggest the wisdom of investing in analytics for any organization seeking to improve performance. Furthermore, our research confirms that while relatively few companies have adopted analytics as a competitive capability, many more aspire to do so. The leaders of these companies have made the commitment to investing in analytics as a means of improving business performance.

Analytics as a Source of Competitive Advantage

Skeptics may scoff that analytics can’t provide a sustainable competitive advantage, because any single insight or analysis eventually can be adopted by competitors. And it is true that an individual insight may provide only transient benefits. Yield management provided a big boost to American Airlines for a time, for example, but using that process is now just a cost of doing business in the airline industry.

Organizations can take several approaches to gain a competitive advantage with data. Some can collect unique data over time about their customers and prospects that competitors cannot match. Others can organize, standardize, and manipulate data that is available to others in a unique fashion. Still others might develop a proprietary algorithm that leads to better, more insightful analyses on which to make decisions. And some differentiate themselves by embedding analytics into a distinctive business process. Increasingly, both data and analytics are incorporated into innovative products and services.

Regardless of the approach, for companies to sustain a competitive advantage, analytics must be applied judiciously, executed well, and continually renewed. Companies that successfully compete on analytics have analytical capabilities that are:

• Hard to duplicate. It is one thing to copy another company’s IT applications or its products and their related attributes (such as price, placement, or promotion), quite another to replicate processes and culture. For example, other banks have tried to copy Capital One’s strategy of experimentation and testing, but they haven’t been as successful. Banks that have been successful with a similar strategy, such as Barclays in the United Kingdom, have figured out their own route to analytical competition. While Capital One relentlessly seeks new customers, Barclays leverages analytics to increase “share of wallet” by cross-selling to its large customer base.
• Unique. There is no single correct path to follow to become an analytical competitor, and the way every company uses analytics is unique to its strategy and market position. For example, in the gaming industry, Caesars uses analytics to encourage customers to play in a variety of its locations. This makes sense for Caesars, because it has long had its casinos scattered around the United States. But that approach clearly would not be the right one for a single casino, such as Foxwoods Resort Casino in Connecticut. It’s also less appealing for casino impresario Steve Wynn, who has translated his intuitive sense of style and luxury into the destination resorts Encore and the Wynn.
• Capable of adapting to many situations. An analytical organization can cross internal boundaries and apply analytical capabilities in innovative ways. Sprint, for example, easily adapted its analytical expertise in marketing to improve its human capital processes. The company applied its “customer experience life cycle” model to create an analogous “employee experience life cycle” model that helped it optimize employee acquisition and retention.
• Better than the competition. Even in industries where analytical expertise and consistent data are prevalent, some organizations are just better at exploiting information than others. While every financial services firm has access to the consumer risk information from FICO, for example, Capital One has analytical skills and knowledge that enables it to outperform the market by making smarter decisions about potentially risky credit customers. The company’s managers refer to the concept of deaveraging—how can they break apart a category or a metric to get more analytical advantage?
• Renewable. Any competitive advantage needs to be a moving target, with continued improvement and reinvestment. Analytics are particularly well suited to continuous innovation and renewal. Progressive, for example, describes its competitive advantage in terms of the agility it gains through a disciplined analytical approach. By the time competitors notice that Progressive has targeted a new segment—such as older motorcycle drivers—it has captured the market and moved on to the next opportunity. By the time other insurers had adopted its approach to pricing based on credit scores, it was working on its Snapshot pay-as-you-drive system.

One caveat: companies in heavily regulated industries, or in those for which the availability of data is limited, will be constrained from exploiting analytics to the fullest. For example, outside the United States, pharmaceutical firms are prevented from obtaining data about prescriptions from individual physicians. As a result, pharmaceutical marketing activities in other parts of the world are generally much less analytical than those of companies selling in the US market. But in other cases, analytics can permanently and abruptly transform an industry or process. As Moneyball and The Big Short author Michael Lewis points out in talking about investment banking, “The introduction of derivatives and other new financial instruments brought unprecedented levels of complexity and variation to investment firms. The old-school, instinct guys who knew when to buy and when to sell were watching young MBAs—or worse, PhDs from MIT—bring an unprecedented level of analysis and brain power to trading. Within 10 years, the old guard was gone.”²⁰

Analytics in Government

We haven’t written much thus far about analytics in government, because our focus in this book is on how organizations compete—and governmental organizations don’t do that in the conventional sense of the term. The one area in which national governments do compete is war, and it’s probably not surprising that the earliest uses of analytics in government involved national defense. The first computers were developed to calculate things like missile trajectories, and Robert McNamara introduced a broad range of analytical approaches to the military—not always with success—when he was secretary of defense in the 1960s. In the present military environment, analytics are used extensively for military intelligence, including automated analysis of text and voice communications (sometimes to considerable public controversy).

Today, however, analytics are widely used at many levels of government, from local to state to federal. They may not necessarily increase governments’ abilities to compete, but they can certainly make governments substantially more efficient and effective. At the local level, for example, perhaps the most impressive accomplishment from analytics has been the use of crime statistics analysis to deter criminals. In New York City, the CompStat program associates crimes with particular geographical regions of the city and is used to guide decisions about where police officers should be stationed. It is also linked with an effort to push decisions down to the precinct level. CompStat has been widely praised for contributing to the reduction in crime in New York since its inception. However, several other factors changed during the same time, so it is difficult to isolate CompStat’s effects alone.²¹

More recently, the Domain Awareness System (DAS), a joint initiative between the New York Police Department and Microsoft, is now being sold to other cities. The system harnesses the power of big data analytics to solve crimes faster and to prevent terrorist attacks. The amount of sensor and other data collected for DAS purposes is astounding, including:

• Images from nine thousand closed circuit video cameras
• Over 2 billion license plate reads from five hundred readers around the city
• Six hundred fixed and mobile (typically worn by officers) radiation and chemical sensors
• An extensive network of ShotSpotter audio gunshot detectors covering twenty-four square miles
• Speech-to-text data from 54 million 911 calls from citizens.
• The system also can draw from internal NYPD crime records, including 100 million summonses.²²

Elsewhere in the United States, several city police departments are using predictive analytics to fight crime and deploy their personnel where they can have the greatest impact. In Atlanta, aggregate crime declined by 19 percent once a predictive policing solution was adopted.²³ Los Angeles experienced a 33 percent reduction in burglaries and a 21 percent reduction in violent crimes in city neighborhoods using the predictive model.²⁴

Beyond crime prevention, there are many possible applications for analytics at the state level, some of which can amount to substantial savings when pursued effectively. Several states, including Massachusetts, have pursued revenue optimization approaches that have yielded hundreds of millions of dollars. These apply to tax and nontax payments. States have also pursued fraud detection to reduce improper payments for welfare, purchase cards, Medicare, and Medicaid. State departments of natural resources have used analytical approaches to model and optimize resources such as minerals, gas and oil, and parks.

Taxpayer compliance was among the US federal government’s earliest nondefense applications of analytics. The Internal Revenue Service (IRS) initiated the Taxpayer Compliance Measurement Program in 1963 to analyze which taxpayers were likely to be cheating on their taxes and to close the “tax gap” between what is paid and what should be.²⁵ It was an effective program for the IRS, but data gathering was judged too expensive and invasive and was discontinued in 1988. The IRS rekindled its analytics efforts with the National Research Program in 2000, and continues to use it as a basis for analyzing compliance and for identifying returns to audit.

One of the most important applications of analytics in government involves health care. This is a major expense for the federal government—its largest nondefense category of spending. Medicare and Medicaid are paid for by the US government but are administered by states. A large medical program that is run at the federal level is the Department of Veterans Affairs (VA) hospitals. The VA has employed electronic medical records and analytics based on them to become one of the most effective health care providers in the United States. As a BusinessWeek article about the VA hospitals titled “The Best Medical Care in the U.S.” described, “In the mid-1990s, Dr. Kenneth W. Kizer, then the VA’s Health Under Secretary, installed the most extensive electronic medical records system in the U.S. Kizer also decentralized decision-making, closed underused hospitals, reallocated resources, and most critically, instituted a culture of accountability and quality measurements.”²⁶

The VA hospitals employ such analytical approaches as predictive modeling of chronic disease, evidence-based medicine, automated decisions for treatment protocols and drug prescriptions, and many others. The VA’s experience is one of the best indications that analytics can have as much of an effect on government as on the private sector.

While the VA hospitals have received criticism about the time veterans have had to wait for appointments, its standard of care has generally remained high. In 2014, Robert McDonald, the former CEO of Procter & Gamble and a strong analytics advocate, was named Secretary of Veterans Affairs. Among other data-driven initiatives, he established a central Data Analytics Division to make analytical expertise available throughout the VA.

Governments around the world are increasingly adopting predictive analytics. Singapore has an integrated city planning initiative called “Smart Nation” that is transforming how public policy decisions are made about issues ranging from strategic (e.g., economic planning) to operational (e.g., transportation planning) to tactical (e.g., determining which books to stock at local libraries).²⁷ Government agencies such as the Irish Tax and Customs Authority use predictive models to target fraudulent tax returns and thereby increase revenues.²⁸ Police forces in the United Kingdom, Singapore, the Netherlands, Uruguay, Brazil, Chile, and the United Arab Emirates are beginning to use predictive analytics to solve and deter criminal activity.²⁹ Emergency services personnel in the Philippines use predictive analytics to improve their preparedness for natural disasters.³⁰

Serving the Market for Analytical Products and Services

While our focus in this book is on how companies can use analytics to optimize key business processes, there is another way to profit from analytics: by making available analytical products or services directly to customers—either as a stand-alone offering or to augment other products and services. For companies in a position to consider this an option, we take a brief detour here.

This focus on data products (which also invariably include analytics) came of age during the Analytics 2.0 era, as we mentioned in the introduction. Perhaps the best-known company of that era that augments its offerings with analytics is Google. In addition to providing analytics for search, which we discuss in chapter 4, and to advertisers, which we describe in chapter 5, Google acquired a web analytics business in 2005 and now makes its Google Analytics service available to anyone, providing data and analytical tools that can be used for search engine optimization (SEO), improving customer engagement, increasing click-throughs, and other marketing initiatives. The company offers Google Analytics with a unique business model: it gives many services away for free (although it has a “premium” version with more sophisticated analytics). Google’s goal in offering the analytical service is to improve the understanding of the web and the internet by providing metrics on website results and user behavior. Rather than competing with other web analytics vendors, Google seeks to “lift all boats” and educate web publishers and advertisers on whether their efforts are working and how to profit from this relatively new channel. The more people who use web analytics well, the better the overall web experience will be, and Google will profit in the long run. Google even provides an online “Analytics Academy” to teach the basic principles of web analytics, in addition to publishing a blog and online articles, offering webinars, and providing public speakers on web analytics.

As we mentioned in the introduction, large firms have also begun to create data and analytics-based products and services. GE, for example, has made a multibillion-dollar bet on the “industrial internet,” including such applications as predictive maintenance for turbines, jet engines, and locomotives; route optimization for locomotives; and clinical analytics in health care.³¹ Verizon Wireless will analyze location data from its customers’ mobile phones on behalf of sports, entertainment, and lodging companies to allow them to better target their ads and offers.³² Philips offers a predictive analytics-based service called CareSage that lets health systems monitor and care for elderly patients by compiling data from wearable devices and home monitors (its own and those from competitors). It identifies patients most likely to have health issues so that clinicians can intervene before problems or hospitalizations occur. In short, if your organization generates data that is relevant to customers, you might want to figure out how best to analyze it and make it available to those customers in meaningful products and services.

Financial investing firms, of course, were among the earliest to employ analytics as a core element of their services. Algorithmic trading is used throughout the industry, and the general feeling is that the decisions in that sector involve too much data and need to happen too quickly to be done by humans. Hedge funds like Bridgewater Associates, Renaissance Technologies, Two Sigma, and others rely on analytics to make almost all trading decisions, and they have done well as a result. As one recent article put it: “Quant funds like those managed by Two Sigma and Renaissance Technologies have been consistently posting solid returns in recent years while most other hedge fund strategies centered around the trading decisions of human beings have struggled mightily.”³³ The leaders of these hedge funds are among the highest-compensated businesspeople in the world, often making billions of dollars per year.

The move to analytics has also been reflected in the rise of analytical consulting. Accenture, Deloitte, IBM Global Services, and many other firms have identified analytical consulting as a growth area in response to client demands. These firms’ consultants help clients address the broad strategic issues around building an analytical capability, or provide assistance building and supporting a particular business initiative (e.g., a customer loyalty program). They frequently tailor solutions for specific industries, such as automated underwriting in financial services. Quantitatively oriented consultants specialize in analytically intensive business solutions, such as supply chain optimization, while specialists in information management help clients develop a robust analytical technical environment. Consultants with particular industry and functional specialties (e.g., marketing or supply chain) work closely with clients while those with purely technical or statistical skills are increasingly based offshore, particularly in India.

Sometimes firms in analytical businesses take on related tasks, such as data management and consulting. In the retail sector, Dunnhumby (which describes itself as a “customer science” company using data and science to help retailers delight customers and build loyalty) worked closely with Tesco on the development of the giant grocery retailer’s loyalty program Clubcard, an important tool in that firm’s ability to use data to optimize its product mix (for more on Tesco, see chapter 5). The company has also worked closely with Kroger in the United States on similar loyalty programs.

Catalina also sells analytical services to the grocery industry that help it understand the effects of coupons and other promotions. The firm retrieves over 300 million transactions per week from more than fifty thousand retailers.³⁴ On behalf of those stores, Catalina manages one of the largest databases in the world, containing the purchase histories of over 260 million shoppers.³⁵ The company aggregates information about purchase behavior and customer demographics, attitudes, and preferences and sells that information to grocery store chains, in addition to offering personalized print and mobile coupons based on analytics. Catalina claims that its approach can increase a company’s average coupon-redemption rates up to ten times higher than they would be with traditional promotion methods.

In many cases, selling data is not enough. Companies also need help interpreting and using the data, and hence buy consulting from external providers. Information Resources Inc. (IRI), for example, has long gathered data from retailers’ point-of-sale terminals, from panels of almost 100,000 consumers, and from pantry audits to understand consumption patterns. More recently, however, IRI has grown its ability to help firms in the consumer packaged goods, retail, and pharmaceutical sectors analyze data to make smarter and more profitable marketing decisions. Sunil Garga, who was President of IRI’s Global Business & Consumer Insights, argues that because of the rise of analytics, “Marketing has changed more in the last twenty-four months than it has in the last twenty-four years, and that rate of change will continue. It’s an analytic revolution.”³⁶

In the case of firms that sell data and analytics, the challenge is often to convince customers of the need for analytical capabilities. According to the managers we interviewed in these firms, a lack of understanding of the analytical methods and what they can accomplish, not cost, is the chief objection. This is one of the reasons why Google has taken such an evangelical and educational approach to web analytics with its Google Analytics offering.

In a variation of the “selling analytics” approach, many established companies are finding innovative ways to enhance physical products by incorporating analytics. For example, medical products companies are designing products with sensors so that an individual’s health data can be analyzed remotely rather than at a clinic or hospital. Also on the horizon are copiers that can transmit data that allows the service provider to schedule preventive maintenance and avoid downtime. And washing machines in the near future will be able to “listen” to sensors embedded inside the clothes in order to determine the right temperature setting.

With the availability of wearable sensors, analytical products are even coming to golf, a sport that has not been a stranger to technological innovation in the past few decades. The Garmin TruSwing consists of a sensor that can be attached to any club and analytical software that runs on smartwatches, smartphones, and tablets. It measures and analyzes metrics based on the mechanics of a golfer’s swing (such as ball flight, trajectory, and distance) to provide insights into each shot and improve performance. Golfers can get instant swing feedback from their wrist after each swing or use a mobile app that allows detailed 3D animations and more detailed analytics. Golfers not only have the ability to analyze individual swings: two swings can be overlaid on top of each other or, alternatively, run side-by-side for further comparison and analysis. Golfers also have the ability to share their performance data, either with friends or their golf pro, to get guidance on how to improve their swing.

When Analytics Aren’t Enough

We wish we could argue that using analytics well is all an organization needs to improve business performance. But there are good examples that disprove that assertion. Large US airlines, such as American and United, are exhibit A. They are analytical competitors but struggled for many years nonetheless. Both airlines (American a bit more than United) were pioneers in adopting such analytical approaches as yield management for seat pricing, optimization of routes and resource scheduling, and the analysis of loyalty program data. While there’s no doubt that these companies would be worse off without their use of analytics, both fared badly during much of the last two decades (although lower fuel prices and industry consolidation have helped them do much better recently, and they’re both hard at work on new analytics approaches).

What happened? Two things kept these firms from succeeding with their analytical strategies over a couple of decades. One is that their analytics supported an obsolete business model. They pioneered analytics for yield management, but other airlines with lower costs could still offer lower prices (on average, if not for a particular seat). They pioneered analytics for complex optimization of routes with many different airplane types, but competitors such as Southwest saved both money and complexity by using only one type of plane. They pioneered loyalty programs and promotions based on data analysis, but their customer service was so indifferent and resources so constrained that loyalty to these airlines was difficult for frequent flyers.

The other problem with their analytical approaches was that other airlines adopted them. Even discount carriers such as Southwest and JetBlue make diligent use of yield management and crew-scheduling analytics. If the discounters lacked internal capabilities for analytics, they could buy them from providers such as Navitaire, PROS, or Sabre Airline Solutions (which used to be part of American but has now been spun out as a separate company). Industry data is widely available from associations and external providers to any airline that wants to analyze it.

In short, there are few barriers preventing any airline from employing standard analytical approaches, and airlines must work very hard to distinguish themselves in analytical competition at this point. Perhaps other frontiers of analytical orientation will emerge in that industry in the future.

Conclusion

The success of companies like Amazon, Capital One, Marriott, Progressive, and Google demonstrates that the use of analytics can lead to better business performance and, indeed, competitive advantage. Over a decade of research indicates that individual analytical projects pay major dividends, and survey data confirms that analytical approaches are correlated with high performance. We have also identified five factors that make an analytical approach a source of competitive advantage. In the next two chapters, we’ll explore in more detail how certain companies are using analytics to outperform the competition. Chapter 4 addresses internal processes, and chapter 5 deals with external processes, such as those involving customers and suppliers.