Framing Unstructured Data for Business Analytics

Where We Are Today

Built upon over 20 years, and with billions of dollars invested, enterprise computing has revolutionized the way companies relate to their customers, the supply chain landscape and how resources are allocated. Organizations now depend on transactional systems, and create summaries from these systems, for decision-making. In fact, regardless of sector, 89% of the highest performing companies say that data is crucial to making strategic decisions (The Economist Intelligence Unit, 2011 Report.). And yet, on the whole, across all surveys, roughly 75% have not tapped into half or more of the valuable data they have. What is keeping these less than stellar performers from realizing the value of their data? The answer: they are operating in silos and not applying business analytics.

Why Business Analytics?

Business analytics provide new information to established processes that support value-based decisions. Analytics focus on answering a specific question, such as which customers are most likely to defect, or how many product units need to be provided. All business analytics share the following characteristics:¹

• They empower action by supporting decisions with data (as opposed to opinion);
• They help anticipate opportunity, with the application of mathematical techniques to transform and summarize input data; and
• They drive impact by adding new value to the original data.

Business analytics provide a learning environment for evaluating results and adjusting activities, data inputs or analysis accordingly. Trends can be examined relative to a baseline, and adapted to new needs, demands or responses.

Staples, the world's largest office products company, wanted to ensure success of nearly 1,500 multichannel marketing campaigns annually. "We did a financial analysis of the implementation, and we found that we were getting an internal rate of return of 137%," says Jim Foreman, Staples' director of circulation and analytics. "That's about as much of a slam dunk as you're going to see."

The practice of business analytics begins with a well-defined plan that describes the problem to be solved, what inputs are available, what the outputs need to address and how the results will be implemented. And done well, business analytics drives top-line opportunities and revenue and lowers bottom-line costs, losses and outages. Because it is an ongoing process, alternate scenarios can be tested, measured and modified to address the dynamics of business environments, supporting innovation and growth.

It "inspires creative analysis," says the US vice president and general manager of Expedia, the world's largest online travel company. Expedia is optimizing its marketing spend, increasing customer lifetime value and improving the overall experience of its online customers. Improving the experience recently helped the company avoid millions of dollars in reservation losses.

Data Is An Asset

In the business analytics environment, data is an asset, and the organizational culture is based on fact-based decisions. But as with any asset, the value depreciates. Data become less valuable to understanding current situations and predicting future conditions because:

• The longer they age they become increasingly irrelevant to existing conditions, forcing reactive responses, while more timely competitors leap ahead;
• Irrelevant noise dampens or confuses the signal, making  it harder to find and understand what is needed; and
• The lack of attention to quality and governance limits the viability of the data.

The fast-forward button has been pressed and the deluge has begun. With an additional 2.5 billion gigabytes of information being produced daily, the transformation of data to decisions requires faster processing. In fact, 40% of high-performance businesses say they have increased their data processing speeds in the last 12 months².

The depreciation of data assets is escalating, with speeds of information encouraged by the proliferation of devices such as the mobile Web and smartphones. The ability to quickly communicate with consumers, suppliers and staff defines the window of advantage held by you over your competitors.

Not all the "big data" are relevant nor lend value to actions. Deciphering what is relevant from Hadoop file systems or other massive data stores demands examination of the data viewed through a "business issue lens" to know what is worthwhile to continue to manage and govern and what can be filtered out.

Unstructured data, which constitutes 90% of the digital universe, provides the verification, corroboration and information to fill knowledge gaps left by structured data. In many cases, the text data is the source of trusted information. In patient records, legal judgments, service notes, call details and all forms of social media, the text provides the rationale behind the action. These are seldom correctly identified in structured fields; in fact, more often than not, structured fields and tags are selected by system end users based on ease of use (what is closest to the top of the menu), versus the meaning they hold to the activity at hand.

What Makes it Big?

Data becomes "big data" when you've exceeded the processing capacity and capability of existing systems. Size alone doesn't make the data "big," (although the influx of social media chatter, scanner and sensor streams, not to mention SharePoint sites profilerating throughout most organizations, doesn't help). Text data is substantially adding to volumes of incoming inputs. Data can be "big" simply because you can't do what you want to do with it. You may be limited to small samples that you extrapolate to the full collection. Or you might be dividing up your business analytics into batch cycles because the lights in the building would dim if you ran it during peak business hours.