BCG Henderson Institute

Generic filters

This is the fifth article in a multipart series.

A few years ago, the idea that data was the new oil caught on. As consumers revealed more of their behaviors, preferences, and attitudes via their electronic devices, companies realized that data, too, is a plentiful, tradeable, and highly valuable commodity.

But data has distinct characteristics that make it very different from oil—or any other commodity. For one thing, the array of sources and amount of supply are virtually infinite. For another, data can be used (or consumed) more than once, and it can be used in multiple places by multiple parties at the same time. Moreover, while it’s hard to hide a tanker full of oil, it’s easy to mask a few billion bytes of data—and then put them to uses (beneficial or nefarious) without the awareness of the original data generators. Data is also unlike most commodities in that different types can have different value, and sometimes that value is not immediately clear. Finally, data can be used for lots of different purposes, many of which were neither contemplated nor intended when the data was first generated. We call these “alternative data uses.”

The myriad alternative uses of data, the ease with which it can be replicated and shared indefinitely at no cost, and the trillions of bytes of data coming on-stream from the Internet of Things (IoT) pose big and far-reaching questions with respect to ownership, privacy, and value. B2B enterprise data sharing, in particular, is just starting to take these issues into account. As companies prospect for new sources of value, the rules, standards, and conventions governing data ownership rights and the regulatory frameworks for privacy and data sharing have yet to take shape.

In this article, we take a tour of the alternative data use landscape and offer some thoughts for business executives who want to realize value from what will be the dominant resource of the 21st century. How should companies think about use cases that are unknown or do not yet exist? How can they balance the abstract value of future use cases with the actual risk of data misuse?

Sources & Notes