Get started with Big Data: tie strategy to performance
October 2012
Dominic Barton, David Court
Large-scale data gathering and analytics are quickly becoming a new frontier of competitive differentiation. In a recent Harvard Business Review article we explore how companies require three mutually supportive capabilities to fully exploit data and analytics: an ability to identify and manage multiple sources of data, the capacity to build advanced analytic models, and the critical management muscle to transform the organization. Getting started on a successful data and analytics journey, however, is a continuing challenge for many leaders and they often struggle for a clear strategy that ties data and analytics to improved performance. We took a close look at companies that have recently launched big data strategies to shed further light on the tough road C-level executives face. From these experiences, we have distilled four principles to defining a strategy and getting started: 1. Size the opportunities and threats Opportunities may range from improving core operations to creating new lines of business — even in the same industry. For example, insurance companies can use big data to improve underwriting performance now, while over the longer term they can use it to serve formerly unprofitable customers and ultimately even develop entirely new risk-based businesses. The key is to establish a clear-eyed view of the business impact expected at each stage of implementation in order to better focus efforts, and determine priorities. In the case of a retailer we studied, data and analytics were part of a difficult battle for market share. The company's strategy had long been predicated on matching the moves of an efficient big-box rival, yet now a different online player was draining the company's revenues and denting its margins. At the heart of the threat was the new competitor's ability to gather and analyze consumer data to generate recommendations across millions of customers while becoming a platform where vendors could sell excess inventory at a discount by using publiclyavailable price data. Resp