Unstructured data resides everywhere in the enterprise and in many places outside the enterprise. It can be in the form of text, video, voice, or graphics; it can be contained in event or message data, rich media data, or within social media. What all of these types of unstructured data have in common is that they hold the key to unlocking latent knowledge capital.
Unlike structured data, which can be handled at scale by many types of technologies, unstructured data presents a seemingly insurmountable problem: much like knowledge itself, it cannot be contained and it requires human interaction to fully realize its potential.
To set the stage for understanding how to unlock the insight in unstructured data, we must begin with a look at the nature of data:
1. Data constantly evolves and changes, in effect replacing itself over and over, as well as being newly created. While it may be interesting to look at historic data for specific areas of your business (as the past can be a predictor of the future), in light of the constant growth and evolution of data, combined with the breakneck speed of business today, it may be more critical to gain an accurate picture of the data as it exists today, in this moment, in order to take immediate action.
2. Data exists in siloes. It exists where it was created, and often, where it is moved to once it had been created (and where it often lacks updating). It is siloed by department within organizations, by social media platforms outside of organizations, by organizations themselves, by geography (within multinational companies), and by subject matter experts (individuals).
Precisely because unstructured data has these characteristics, constantly changing and growing and existing in siloes, it cannot be stored efficiently over time and still provide the type of insight most companies need to transform their performance. What exists today may be gone tomorrow, replaced, tweaked, or just outdated by entirely new data. What exists in siloes cannot be combined with other data, or, in many cases, even found by those outside the silo. This is why it is critical to not only allow data to exist where it naturally resides (so it remains fresh), but also to connect the data in real time to enable actionable insight.
How is this accomplished?
Monitor. Listening can provide certain benefits, especially within social media. However, listening without responding negates the value of the data. Many organizations use text analytics on individual social media streams and communities, which is a good first step (Zach Hofer-Shall, an analyst with Forrester who covers customer intelligence, refers to this step as "crawling" in his report Roadmap to Integrating Social and Customer Data.) This analysis generally creates more data around a single data stream (the social platform), which is subject to data characteristic #2, it is siloed, generally within marketing or customer support, and also #1, it grows old and requires updating. This first step is generally inadequate on its own because it does not relate the social data (or other data that is being monitored, such as phone calls or chat logs in the call center) to other data that will lend it meaning and provide the insight to take action.
Integrate. Organizations can most effectively monitor and analyze-and take action based on-unstructured data across multiple channels simultaneously-but only if it's connected. Historically, this would have required system integration-which consumes too many resources and takes too much time. There is another, faster and more optimal way, particularly considering the transient nature of data. An index of data-think about it as a virtual integration-is today's best practice for efficient and non-disruptive data connection.
Analyze. Performing multichannel text analytics on the unified index enables relationships to be formed among the unstructured data, in real time, to provide a much more accurate picture of what is happening now and to provide the key information for decision-making and action: Is this a trend? Should we respond to this, and if so, how? Is it an individual issue and, so, less relevant? Or is it an individual issue related to one of our most valued customers? Such decisions require insight, available only by exposing consolidated and correlated knowledge, to human interpretation. In these examples the organization would have integrated and analyzed unstructured data from social media and customer communities alongside data from CRM systems, product defect tracking systems, email communication, chat, voice, and more.
As important, the unstructured data must be presented in interfaces that facilitate human interaction: allowing employees, customers, and partners to navigate and make sense of it; in effect, to discover new ideas, new relationships, and new ways to drive business, all thanks to its unlocking the latent knowledge capital. Today, there is a new category of software being established: insight solutions. These solutions are aimed at mining the mountains of primarily unstructured data to support business processes and presenting it in ways that enable humans to unlock its potential. And with this comes the ability to truly unleash the power of your data, no matter where it resides.