5 Ways Your Data Strategy Can Fail

Executive Summary

It takes a lot to succeed with data. For many organizations, progress is incredibly slow. And while data is being put into more processes, it is rarely seen on the balance sheet and income statement. In order for data efforts to be successful, a company must perform solid work on five components, each reasonably aligned with the other four. First, companies need properly defined, relevant, structured, and high-quality data. Second, companies need a means to monetize that data. Third, organization capabilities must be fit for data in three arenas: talent, structure, and culture. Fourth, companies need technologies to deliver at scale and low cost. The last element is defense, essentially minimizing risk.

Mint Images/Getty Images

There are plenty of great ideas and techniques in the data space: from analytics to machine learning to data-driven decision making to improving data quality. Some of these ideas that have been around for a long time and are fully vetted, proving themselves again and again. Others have enjoyed wide socialization in the business, popular, and technical press. Indeed, The Economist proclaimed that data are now “the world’s most valuable asset.”

With all these success stories and such a heady reputation, one might expect to see companies trumpeting sustained revenue growth, permanent reductions in cost structures, dramatic improvements in customer satisfaction, and other benefits.  Except for very few, this hasn’t happened. Paradoxically, “data” appear everywhere but on the balance sheet and income statement. Indeed, the cold reality is that for most, progress is agonizingly slow.

It takes a lot to succeed with data. As the figure below depicts, a company must perform solid work on five components, each reasonably aligned with the other four. Missing any of these elements compromises the total effort.

Let’s explore each component in turn.

Quite obviously, to succeed in the data space, companies need data, properly defined, relevant to the tasks at hand, structured such that it is easy to find and understand, and of high-enough quality that it can be trusted. It helps if some of the data are “proprietary,” meaning that you have sole ownership of or access to them.

For most companies, data is a real problem. The data is scattered in silos — stuck in departmental systems that don’t talk well with one another, the quality is poor, and the associated costs are high. Bad data makes it nearly impossible to become data-driven and adds enormous uncertainty to technological progress, including machine learning and digitization.

Then, companies need a means to monetize that data, essentially a business model for putting the data to work, at profit. This is where selling the data directly, building it into products and services, using it as input for analytics, and making better decisions come to the fore. There are so many ways to put data to work that it is hard to select the best ones. A high-level direction such as “using analytics wherever possible” is not enough. You have to define how you plan to use analytics to create business advantage and then execute. Without a clear, top-down business direction, people, teams, and entire departments go off on their own. There is lots of activity but little sustained benefit.

Insight Center

Organizational capabilities include talent, structure, and culture. Some years ago, I noted that most organizations were singularly “unfit for data.” They lack the talent they need, they assign the wrong people to deal with quality, organizational silos make data sharing difficult, and while they may claim that “data is our most important asset,” they don’t treat it that way. If anything, this problem has grown more acute.

Start with talent. It is obvious enough that if you want to push the frontiers of machine learning, you need a few world-class data scientists. Less obvious is the need for people who can rationalize business processes, build predictive models into them, and integrate the new technologies into the old. More generally, it is easy to bewail the shortage of top-flight technical talent, but just as important are skills up and down the organization chart, the management ability to pull it all together, and the leadership to drive execution at scale. Consider this example: Many companies see enormous potential in data-driven decision making. But to pursue such an objective, you have to teach people how to use data effectively (HBR’s current series on data skills will focus on this topic). Leadership must realize that earning even a fraction of the value data offer takes more than simply bolting an AI program into one department or asking IT to digitize operations.

Structure and culture are also a concern. As noted, organizational silos make it difficult to share data, effectively limiting the scope of the effort. All organizations claim that they value data, but their leaders are hard-pressed to answer basic questions such as, “Which data is most important?” “How do you plan to make money from your data?” or “Do you have anything that is proprietary?” Some even refer to data as “exhaust” — the antithesis of a valued asset! Without an abundance of talent and an organizational structure and culture that value data, it is difficult for companies to grow successful efforts beyond the team and department levels.

Fourth, companies need technologies to deliver at scale and low cost. Here, I include both basic storage, processing, and communications technologies, as well as the more sophisticated architectures, analysis tools, and cognitive technologies that are the engines of monetization.

Quite obviously companies need technology — you simply can’t scale and deliver without it. Facebook, Amazon, Netflix, and Google, who have succeeded with data, have built powerful platforms. Perhaps for these reasons, most companies begin their forays into the data space with technology. But from my vantage point, too many companies expect too much of technology, falling into the trap of viewing it as the primary driver of success. Technology is only one component.

The last element is defense, essentially minimizing risk. Defense includes actions such as following the law and regulations, keeping valued data safe from loss or theft, meeting privacy requirements, maintaining relationships with customers, matching the moves of a nimble competitor, staying in front of a better-funded behemoth, and steering clear of legal and regulatory actions that stem from monopoly power. You’re unlikely to make much money from defense, but poor defense can cost you a lot of time, money, and trouble.

Thus, data require a range of concerted effort. At a minimum, HR must find new talent and train everyone in the organization, tech departments must bring in new technologies and integrate them into existing infrastructures, privacy and security professionals must develop new policies and reach deep into the organization to enforce them, line organizations must deal with incredible disruption, everyone must contribute to data quality efforts, and leaders must set off in new, unfamiliar directions.  Adding to complications, data, technology, and people are very different sorts of assets, requiring different management styles. It’s a challenging transition. Many companies have tried to resolve their data quality issues with the latest technology as a shortcut (e.g., enterprise systems, data warehouses, cloud, blockchain), but these new systems have missed the mark.

It is important to remember that the goal is not simply to get all you can out of your data. Rather, you want to leverage your data in ways that create new growth, cut waste, increase customer satisfaction, or otherwise improve company performance. And “data” may present your best chance of achieving such goals. Successful data programs require concerted, sustained, properly-informed, and coordinated effort.