Are You Ignoring Unstructured Organizational Data?

Jigsaw Pile

In this Point of View, Adam Rogers, co-CEO and chief technology officer at Ultimate Software, shows why unstructured data can drive game-changing use of advanced technologies.

Decision makers love data.

But until recently, organizations have primarily relied on structured data—highly organized data sets that are easy to analyze. Unstructured data, like emails and text, don’t usually follow a predefined model or fit into relational databases. Therefore, a great deal of enterprise data is unstructured, but many organizations may ignore it completely because of the inherent difficulties of analyzing it.

Why make decisions based on incomplete information? Unstructured data can be game-changing to your business. Here's why. @UltimateHCM @AdamR #HRTech #HRTribe #HR Click To Tweet

Can you imagine making crucial business decisions based on only some relevant information? In addition to the sheer volume of unstructured data, I find it particularly valuable due to its ability to provide rich, detailed and qualitative insight into what’s truly happening within your organization—not just the what, but also the why.

I’ve dedicated my career to tech innovation at a cloud provider of Human Capital Management solutions. We leverage unstructured data analysis and relational learning to drive prescriptive-analytics solutions and sentiment-analysis capabilities for human resources using natural language processing (NLP) models.

I believe advances in NLP, pattern recognition and cognitive analytics are absolute game-changers that can empower businesses across industries to collect and make use of unstructured insights. From customer acquisition to retention and everything in between, this technology has endless practical applications.

Mining Data To Drive Sales And Retain Customers

Ultimate Software's Adam Rogers
Ultimate Software’s
Adam Rogers

Marketing and customer success teams, for example, rely on various hard-data sources–such as engagement and conversion rates–to prove campaign success. While these data are certainly valuable, they’re unable to explain why a customer or prospect behaves in a certain way. To answer that, you likely need unstructured data.

Already, we’re seeing the emergence of artificial intelligence (AI) solutions that couple with CRM systems like Komiko to provide account managers insight from unstructured email and calendar data. These platforms can perform a multitude of tasks, from tracking and documenting real-time communications to leveraging AI tools like Conversica for finding new customers that automate follow-up and flag warm leads–a tool explored in a Wall Street Journal article.

Looking to the future with the incredible explosion of business-to-business (B2B) online review sites, I expect we’ll start seeing solutions emerge that can not only crawl the web but also simultaneously capture sentiment, which could allow organizations to easily monitor and improve both customer satisfaction and online reputation.

Creating A New Frontier in Patient Care

Already, healthcare organizations around the world are leveraging AI technology to vastly improve patient care and outcomes.

For example, according to a Science Daily news release, researchers from Imperial College London developed an AI solution capable of determining the best treatment strategy for patients with sepsis. After analyzing extensive records of approximately 100,000 intensive care unit patients, including 48 structured data sets (such as age, blood pressure and heart rate), the technology learned to apply previous experience to diagnose the best course of treatment.

Ninety-eight percent of the time, the AI system reportedly either “matched or was better than the human doctors’ decision.” And when doctors deviated from AI recommendations, patients had a “reduced chance of survival.”

Attempts in other locations have had similar results. West Virginia’s Cabell Huntington Hospital reportedly developed a separate machine learning solution to address sepsis through early detection and clinician alerts, and its sepsis-related mortality rates dropped 33.5%. In California, a small randomized controlled clinical trial found similar success in decreased length of stays using a predictive algorithm.

However, in these and many other successful healthcare case studies, the systems are limited to structured data. Consider the role text, image, audio and video data play in most diagnoses and the fact that each of these data sets are typically analyzed separately by human beings. What if we could train technology to effectively analyze these unstructured data points for us?

Flatiron Health, which was reportedly acquired for $1.9 billion, is working on this—but it’s not there yet. Flatiron’s oncologist-facing platform, as explained by one Forbes contributor, has differentiated itself with a dataset brimming with unstructured oncology data and features like free-text pathology reports and clinical notes. In cancer diagnoses in particular, these unstructured insights could be crucial.

But while Flatiron’s solution is driven by tech, it appears limited in its ability to actually extract insights from unstructured data. For the time being, Flatiron’s solution is reportedly paired with healthcare professionals who manually analyze and contextualize the data.

As NLP functionality and neural network algorithms improve, I expect these limitations will fade away. The widespread use of unstructured data analysis in health care could open the door for more accurate diagnoses (including early detection), enhanced treatment quality, faster turnaround times, and fewer specialist referrals, essentially serving as a catalyst for more efficient and affordable health care models.

Developing An Unstructured Data Strategy

With so many emerging opportunities to leverage unstructured data, I believe carefully developing data strategies and frameworks should be a top priority for executives.

A vital first step is to decide which sources of data will have the biggest impact on your business and start there. I’ve found that homing in on a few crucial metrics rather than overloading your team with too many priorities almost always produces better results.

Once you’ve identified your needs and goals, spend some time researching existing software tools. Depending on your situation, it may be more efficient to partner with a vendor than to develop a new solution in-house. Outsourcing often results in better resources, compliance, flexibility and expertise; on the flip side, doing the development in-house usually means better privacy, increased employee goodwill and more control over communication and management.

Finally, consider your culture. If there’s a disconnect between how the data tools are designed and how managers will use or apply them, long-term success is unlikely. Ensure that whatever model you use will complement existing processes, that resulting analytics can be easily understood by frontline managers and that—perhaps most importantly—you’ve nurtured a data-driven organization that encourages and rewards proven methodologies and results.

Adam Rogers is co-CEO of Ultimate Software, a leading cloud provider of HCM solutions. Based in Weston, Fla., Ultimate Software is a sponsor of the HCM Technology Report. To learn more, click here. A version of this article was originally published in Forbes.

Sign up for our newsletter here.

Image: iStock

Previous articleUltimate Software Releases New Report on State of Remote Work
Next articleCapacity, Nee Jane.ai, Unveils Knowledge Sharing Platform