Analytics professionals need greater awareness of when analytical tools should not be used.
Organizations no longer ask themselves “Could we do X with data?” The answer is now often yes. Instead, a key question now is, “Should we do X with data?”
With analytics as a hammer, so many questions can start to look like nails. It is difficult for organizations to know what to do. But the “should” in “What should we do?” goes beyond just selecting what to hammer on for maximum insight — the possibilities that analytical abilities create involve responsibilities as well.
Questions involving “could” are about ability. Can we estimate which customers are unlikely to renew their contracts? Can we determine people who are about to make large financial decisions such as real estate purchase or retirement? Can we figure out which machine part is mostly likely to break so that we can order a replacement ahead of time?
Problems like these are difficult; they used to be largely relegated to experience and were particularly difficult to investigate with analytical approaches. Historically, several bottlenecks restricted organizational use of data and analytics. For example:
- Data was insufficient to support decision making; it fell short on multiple dimensions such as quality, quantity, timeliness, etc.
- Technologies to store, transform, and analyze data were expensive and/or complex.
- Talent skilled in analyzing and developing insights from data was not common.
- Organizational culture did not promote analytics as a best practice, treat data as a core asset, or use analytics to guide strategy.
- Isolated data or difficulties aggregating data kept decision making tactical rather than strategic.
But now, prior constraints may no longer hold. Consider:
- Corporations rely on increasingly mature information systems embedded throughout their value chain that, while still far from perfect, provide increasing quality and quantity of data. The data sources now reach outside the organization through interconnected devices; companies like Enernoc bring Internet of Things data into organizations.
- Technologies that were once novelties are now widely available.
- While analytics talent is still not ubiquitous, organizations are finding ways to develop analytical capabilities both from outside and within develop analytical capabilities both from outside and within.
- Organizational cultures are shifting to more data-oriented mindsets.
- Infrastructure improvements and master data management allow enterprise wide reporting, even for the largest organizations. Coca-Cola, for example, is developing a common language about data in order to shift conversations from the data itself to the use of strategic insight from it.
So now what? As organizations are able to do more with data, analytics can answer more and more questions — but not all of them should be asked.
Take Equifax as an example. As an aggregator of vast amounts of data, the company is in a position to know detailed information about the individual lives of most consumers — where we are, what we do, and who we are with. However, just because Equifax can ask these questions, doesn’t mean the company feels it should.
Instead, Greg Jones, vice president of Enterprise Data & Analytics at Equifax, describes the company’s concern about the implications of such detailed analysis. “From an Equifax perspective, we’re not going to risk a consumer’s privacy, and we don’t want to risk our core business. If we’re going to improve a model by a tiny percent, but it’s going to push us closer to that line, we’re not going to do it.” Jones notes that, along with its ability to use data, Equifax has thought carefully about the responsibilities associated with the data. “Our primary focus is on being a trusted steward of data and keeping people’s information secure,” he adds. “Secondarily, it is to use that data in a manner that is legal, responsible, and helps improve consumer’s lives and businesses’ risk.”
Like hammers, analytical tools themselves are agnostic. And, like any tool, analytics can be used responsibly or irresponsibly. Responsible use of data and analytics capabilities requires that organizations think through the implications of the analyses they do. When MedSeek identified potentially profitable patients for hospitals, it raised concerns about fairness. And there is something troubling about the use of obituaries to make a real estate sale. It doesn’t feel right.
As an alternative, organizations need maturity around analytics. The distinction between “could” and “should” is important. As individual lives become increasingly measured, quantified, and subject to analysis, the potential for harm is real. Just like engineering has matured from ad hoc beginnings, organizations need ethical processes around analytics.
Topics like governance are not headline-worthy. But no engineer wants a headline of “bridge collapses,” and the results of analytics that are performed and used (but should not be) could easily become that type of catastrophe.