9 Steps to Unlock Big Data's Hidden Value

No matter where you look, the numbers on big data are staggering. Among estimates of its potential benefits are productivity-led savings: $300 billion a year for the US healthcare industry; €250 billion for the European public sector; a 60 percent potential increase in retailers’ operating margins; and $600 billion in economic surplus for services enabled by personal-location data. However, these are just the early calculations coming in from a few sectors; they could well go higher.

So, What Exactly Is Big Data?

Big data, or big data analytics to be precise, is the ability to extract highly important information, insights and trends from massive amounts of data. But how does it add value to business? Big data helps leaders measure and understand more about their business and directly translate it into improved decision making and performance.

It is being billed as the next big thing and every business appears to be surfing the big wave. But if the hype makes you feel you’ve been left behind, fear not. Gartner poll results released in September 2014 revealed that while 73 percent of firms surveyed in 2014 are interested in adopting big data, only 13 percent have actually deployed these technologies. Without a clear place to start, many organizations are either misusing big data or not using it to its full potential.

However, a focused approach can help make the transition less daunting, minimize the hurdles along the way, and enable organizations to take advantage of this valuable and growing corporate asset. By breaking it down to nine steps, businesses can help unlock the hidden potential of big data.

1. Define Responsibilities

With an overwhelming amount of raw big data, it can be hard to see a clear path forward. By defining responsibilities early on in the data management planning process, organizations can ensure smoother adoption and organization-wide integration. While the responsibility of collecting data should be shared by the IT and analytics teams, the final analysis of big data should be managed by analytics professionals. In fact, senior-level management should be responsible for identifying areas within their respective teams where big data could drive value.

However, sometimes getting this level of support from executive management is not easy, especially if the IT and analytics teams (or a dedicated big data center of excellence) reside outside of their business function. That’s why it is important to assign a point person in each business function who can look for opportunities to use big data to improve business outcomes and foster support of big data analytics internally.

2. Guide Business Functions to Ask the Right Questions

Executives will find it easier to garner buy-in if they can demonstrate how big data analytics can be valuable to the business. A simple question like “What would you really like to know about your business, and how can data help you with it?” is a logical starting point.

Such questions could trigger more exploratory questions by functional experts. For instance: Are our investments in customer service paying off? What is the optimal price for our product right now? Or even: What is the value of a ‘tweet’ or a ‘like’? The ability to ask the right questions is the key to succeeding with big data. It is also important to remember that big data is about discovering insights with data that can lead to valuable outcomes.

3. Take Stock of All Data “Worth Analyzing”

Valuable business insights can come from many sources, including social media, activity streams, “dark data” (unused captured data), machine instrumentation and operational technology feeds. It is important to explore these sources and experiment with new ways of capturing information, such as complex event processing, video search and text analytics. An organization’s data typically fit into four areas:

  • Operational data: smart grid meters data, embedded systems (microwave sensors, chips inserted in automobiles), transactions logs (payment transactions), radio-frequency identification chips (RFID), navigation and location sensors, networks, and servers, etc.
  • Streaming data: computer network data, phone conversations, etc.
  • Documents and content: PDFs, Web content, and legal discovery elements (electronic information exchange in civil litigations), etc.
  • Rich media: audio and video tracks, electronic images, etc.

4. Select the Functions Best Positioned to Lead

It is logical to launch big data initiatives in business functions that are the most prepared to collect and analyze data and will receive the highest potential reward. In our experience with clients, functions poised for maximum growth include marketing, customer service, supply chain management and finance.

5. Match Initiatives With Business Functions

Some big data programs can be implemented in a variety of settings, but most are suited to specific functions. Here are some scenarios that illustrate how different business functions can harness the power of big data:

  • Customer functions (marketing, e-commerce, and customer service): Targeted advertising that provides personalized offers to consumers based on their socio-demographic characteristics; loyalty management that extends channel reach from point of sale, Web, and call center to include mobile and social capabilities.
  • Finance functions (finance, risk, and treasury): Intraday liquidity management by providing real-time monitoring of price movements in relation to positions, to make trading and rebalancing decisions; improved credit risk assessment through multiple big data-supported credit risk assessments that factor in hundreds or even thousands of indicators.
  • Supply chain and procurement functions: Better dynamic route optimization through more iterations and faster route planning in real-time.

6. Find New Value

Making the case for a big data initiative is easiest when it can be shown to create new value. In comparing data views from a traditional business intelligence perspective versus a big data viewpoint, consider the following: What data is being captured? What are the limitations of this kind of structured data? What extra value will be gained through external, context-specific and unstructured data? Where will data be found and how will it be collected? Would our business act upon the insights gained? Is the extra business value worth the additional investment of time, energy and money?

The sheer variety of potential value creation – from clinical trials and marketing to risk management and audits, from analyzing crop and seed production to fan listening posts – is astounding.

7. Assess Complexities and Prioritize

When beginning a big data test, it is helpful to remember the complexity of both the type of data and the type of analysis required. Often, big data comprises unstructured information that has traditionally been impossible to break down or categorize. This makes big data difficult to analyze and easily misinterpreted. Therefore, it’s easiest to begin a program with data that is relatively easy to analyze.

Different types of data analysis also present varying degrees of complexity. Generally speaking, descriptive analytics (answers “what happened?”) are relatively easy, whereas diagnostic analytics (answers “why did it happen?”), predictive analytics (answers “what will happen?”) and prescriptive analytics (answers “how can we make it happen?”) are increasingly complex to conduct.

8. Assess Your Technology Architecture

Many traditional and even state-of-the art technologies were not designed for the volume, velocity and variety of data available today – or expected tomorrow. As data sets grow exponentially, the investments required for scaling technologies to perform efficiently grow even faster. Organizations should consider the various methods to upgrade their infrastructure to support or anticipate the influx of big data. Creating or upgrading to big data-ready technology architecture is no small feat. Building everything from scratch takes time, and buying everything is expensive. Therefore, finding the right combination of insourcing and outsourcing requires careful consideration.

9. Build a Strong Team

Big data is expected to witness a big talent crunch. McKinsey Global Institute estimates that by 2018, the U.S. alone could face a shortage of 140,000 to 190,000 workers with deep analytical skills. That’s not including the 1.5 million managers and analysts with the know-how to use big data to make effective decisions.

Don’t let these statistics scare you away from investing in a strong big data team. You can start today and build your team step-by-step:

  • Break down your talent needs: Consider organizing talent into groups reflecting the four distinct roles – business analysis, analytics, database technology and data visualization.
  • Scan internally for the right skills: Although they may not be in the right department, every organization probably already includes people who know the business, possess data-crunching capabilities and can make data-driven decisions.
  • Hire people with needed or related skills: If the required talent is not available or cannot be acquired by cross-training existing employees, consider adding new team members. If talent with the needed skills is unavailable, consider hiring someone with related skills. For example, organizations should consider substituting statisticians for more scarce data scientists.
  • Start small and scale up: In the beginning, your needs will be modest. A few hires may be adequate to get started with your big data initiatives.

The big data journey has just begun. Many global organizations are embarking on it to derive new insights. Yet, big data can only improve your business if you start making decisions based on its findings. And this will require leaders with a data-driven mindset, those who believe in decisions based on hard numbers rather than gut feel or past experience.
Anant Gupta is CEO of HCL Technologies, a $6.5 billion global provider of IT and engineering services.

This story originally appeared on Information Management.

For reprint and licensing requests for this article, click here.
Analytics Data and information management Core systems
MORE FROM DIGITAL INSURANCE