Why businesses need a strategy for big data

Conferences and journals regularly explore big data’s potential to transform the nature of executive decision-making, or to fundamentally revamp a particular business model. However, despite often large investments, many organisations are still not getting the most from big data. To improve their chances of positive results, and avoid falling behind their competitors, they need to acquire a more calculated approach, adopting a defined set of strategic principles.

Many executives find themselves caught in a big data bind. On the one hand, they may strongly believe, on a theoretical level, that big data could be transformational. On the other hand, these same executives may have already invested heavily in big data projects that have not achieved tangible results. Moreover, project costs look set only to escalate as the amount of potentially invaluable information, both internal and external, grows exponentially.

It is perhaps no surprise, therefore, that risk-averse executives will often baulk at the option of spending large amounts of money on projects which they suspect may fail to achieve their desired objectives. A 2014 Economist Intelligence Unit report found that just a quarter of executives had prioritised the findings of big data when making their latest major decision. This approach inevitably leaves data-shy companies vulnerable to competitors who base decisions on crucial nuggets of data analysis.

To help to repel this competitive threat, and boost the probability of tangible results, organisations need to adhere to several strategic principles which should form the foundation of any approach to Big Data projects.

First and foremost, big data programmes need to be business-driven and not data-driven. Executives need to focus on specific business priorities which can then be translated into precise business objectives. The huge volume of available big data presents opportunities in all sorts of business areas. It is easy to lose sight of where the data can yield the greatest commercial advantage, and spread resources too thinly.

Relentlessly defining objectives before finalising the investment is therefore of paramount importance. Well-intentioned but vague goals such as ‘improving customer satisfaction’ are unlikely to provide the necessary focus. However, an exact and measurable objective such as ‘providing customised products to the top quarter of our customers’ will serve to concentrate efforts on the most relevant data sets and internal analytics capabilities.

Second, decision-makers should seek the most efficient project length. They should be mindful of the sheer pace of technological change, and avoid overly long projects where operating technology can become outdated before completion. Similarly, if the projects are too short, they can be insufficiently rigorous. Companies should aim for projects which are brief enough to prevent technological obsolescence, but long enough to be genuinely effective. Around two years could be the ideal timeframe.

Third, companies should be mindful that data projects don’t have to be enormously expensive and wide-ranging to produce invaluable outcomes. They ignore ‘little data’ at their peril. For example, one regional health insurance company in the US took full transcripts of its calls with customers, and applied text mining algorithms. Simply by means of this relatively minor investment, the company was able to improve the format and language of its written communications, and make the call centre process more efficient.

Fourth, before they make large-scale investments in technical infrastructure and tools, executives must consider carefully what data and analytics are required. This will help them wade through the vast number of available tools on the market, and select the most appropriate. By way of example, in excess of 150 NoSQL databases, a bewildering level of choice, can now be purchased. Different types of job will benefit from different tools, and marrying the two effectively is certainly not a simple process.

Finally, leaders must refine their talent strategy. Big data projects will only be successful if they are staffed by the right people, and are given the appropriate senior-level support. The entire area of data science is in its infancy, and the pool of qualified and able people is very limited. The task of building a sufficiently strong internal big data team with the requisite range of skills and experience may therefore be insurmountable, certainly in the near term. Partnerships with leading academic institutions or service providers may offer the simplest way to bridge that talent gap, introducing the necessary skills into the organisation until such time as traditional recruitment in the marketplace can be relied upon.

Data projects are set to become part of the furniture of organisational life for the foreseeable future. Companies therefore need to get their basic approach right before embarking on an assortment of costly projects. Applying this discipline will enable them to extract the greatest possible competitive advantage from the huge volume of data at their disposal.

Samer Bohsal, Abdulkader Lamaa and Ameep Pandey are partner, principal and associate respectively at Strategy& (formerly Booz & Company).