Digital data strategy starts with business outcomes
CIOs need a new strategy to manage data in the digital era. A good data strategy starts with clear identification of potential business outcomes and works back from there — not just to the technologies and data, but to the people and processes needed to achieve those outcomes.
Starting with business outcomes means that data that has the potential to yield business value gets the most attention. For example, if a company wants to boost its customer share of wallet, then the data that allows the company to estimate the customer’s potential to buy more products would be prioritized over other data about the customer. And if the business outcome changed from share of wallet to capturing new customers, then the focus of the data initiative would change too. Hallmarks of this flexible, business outcome-driven approach include:
1. Actively promoting business-led analytics: While data infrastructure expertise still lies in IT, progressive organizations recognize that the best ideas for using data come from the frontlines. Analytics teams in the rest of the business are increasingly common, but they vary widely in maturity and focus. For instance, some are simply business intelligence reporting groups that have been renamed data science or analytics, while others are highly experienced teams that are fully capable of using advanced analytics.
A company’s data strategy should encourage the development of these analytics teams. IT should work closely with the rest of the business to hire data scientists and analysts into business lines. But hiring isn’t enough. The analytical skills of the existing workforce also need attention. Many companies find that the majority of employees lack the skills and judgment to use data effectively to make decisions. Consequently, an effective data strategy should include efforts to close this analytical skills gap.
2. Making data governance iterative and ownership collaborative: Traditional data governance policies tend to be static, permanent and clear-cut. For example, at many companies, data stewardship and standards take years to bed down and even more years to change. Similarly, data ownership is usually binary — someone either owns the data or they don’t. None of this works well when there are rapid changes to data technology, users or uses. Instead, digital data strategies should include flexible, iterative and collaborative governance models that can quickly capture value from new ways of using data and recognize that those uses often involve handoffs between many parties.
3. Disaggregating data and users: Not all data should be treated the same way, and not all users should be forced to use the same analytics tools. Some types of data and certain groups of users create much more value than others and should be treated differently. As such, data strategies should support a range of tools and let more mature analytics teams choose those that suit their needs. Similarly, strategies should recognize that quality thresholds vary and there are many situations where less-than-perfect data is still valuable.
Additionally, as more analysis is automated, data strategies should not overlook the teams that create algorithms used for machine decision-making. For instance, groups that create sensors on capital equipment that predict when maintenance is required have specific data and technology needs and a high level of maturity. They should not be supported in the same way as an average reporting group.