There’s a general belief that if you want a successful business, you must start collecting data—and lots of it. Data is the new gold, the new oil. After all, if Google and Facebook, two of the world’s most valuable companies, built their empires atop vast amounts of user data, why shouldn’t it work for other companies?
Reality check: Though data can bring immense value to almost any online business, simply collecting it isn’t enough. In fact, too much data can have an adverse effect on your business. Data needs to be mined and turned into something of value.
“Technology is meaningless unless it has an impact, directly or indirectly, on human life,” says Guy Levy-Yurista, VP Strategic Growth and Innovation at business analytics company Sisense. “In the same manner, data is meaningless unless you are able to extract insights that impact human life. The value of those insights is their impact on business and by extension on human life and the human environment.”
Gut Instincts Aren’t Always Best
Companies and executives make two major mistakes when it comes to dealing with data in their businesses. The first is ignoring data and relying on gut instincts to make decisions.
“That usually has to do with managers and executives who grew up in a world that preceded data-driven methodologies,” Levy-Yurista says. “They’re falling back on their experience rather than data, because the world has changed around them, and they failed to notice or acknowledge this new reality.”
The progress of data-driven business tools is slowly forcing decision-makers to adapt or risk falling behind. Big data won’t replace human intuition, but it can augment it.
Don’t Be a Data Hoarder
The second mistake is hoarding too much data. Companies with little or no experience in handling big data tend to look at data-driven companies and conclude that the key is to collect lots of it. But then what? Many companies fail to put the data they collect to meaningful use, because of a lack of in-house tools or skills, or simply because they don’t have a strategy.
“Yes, you do have to collect all the data that you can, but you shouldn’t be drowning in it,” Levy-Yurista says. “You need the right tool to distill that data.”
According to consulting firm McKinsey, the connectivity provided by the IoT industry presents a $11.1 trillion opportunity. But that potential is largely dependent on companies being able to correctly analyze and integrate data into their business processes.
“If organizations cannot manage their vast amounts of data, they cannot take advantage of its value,” says Amnon Drori, CEO of Octopai, which specializes in business intelligence and data management automation.
Scrubbing data, anonymizing it, and making sure company servers store only valuable information is expensive. And collecting too much data without distilling could run afoul of data privacy rules. With the General Data Protection Regulation (GDPR) looming, companies—both European and those that do business with European companies—will have to consider what kind of data they store and whether it contains personally identifiable information (PII).
“Organizations are spending more and more time on tracking, finding, and understanding data and the metadata behind it,” according to Drori, who says some companies Octopai has started working with spend up to 50 percent of their time gathering and processing data.
Removing Technical Barriers
When it comes to analyzing data, some executives and decision-makers are limited to preset dashboards and reports from their business intelligence teams. But artificial intelligence and machine learning are helping companies use their data in meaningful ways.
Machine learning algorithms let software quickly find correlations and recurring patterns in vast amounts of data with little or no assistance from human experts. They also let humans peruse their data repositories in easier and more intuitive ways.
Narrative Science specializes in natural language generation (NLG)—a branch of machine learning that deals with generating human-understandable output from data. It teamed up with business intelligence company Tableau to provide automatically generated written descriptions of analytics charts, so people with minimal experience in data analytics can use it in their work.
IBM Watson, one of the leading platforms in data analytics, also employs artificial intelligence to simplify data management. It uses natural language processing (NLP) to enable users to query their data sources through simple sentences.
Sisense also enables companies to manage complex datasets with its BI platform, and provides AI tools such as NLP-powered chatbots. Users can query their data through simple conversational interfaces, such as Skype or Slack. In a recent upgrade, Sisense added intuitive visual drag-and-drop tools for data preparation, so nontechnical business users can easily find, add, and combine complex data sources to their data platforms.
This approach, Levy-Yurista believes, is beneficial to companies struggling with making productive use of their data as well as to organizations that have yet to adopt a data-driven business model.
Metadata Is a Must
“In order to make the most out of our data, we have to automate metadata management,” Octopai’s Drori says. “Data actually means nothing without metadata.”
Metadata, also called “data about data,” comprises additional data points that help classify otherwise unstructured information. For instance, in a messaging app, each message exchanged between two users could contain metadata such as sender, receiver, the time the message was sent, the IP address and device type of the sender, and more.
“Metadata is the glue that holds all our data together and helps us align data sources so that we can understand our data and use it to make smarter decisions for our business,” Drori says.
One of the problems companies face is that their data often resides in different silos—cloud storage, on-premise storage, and business intelligence platforms, for example. Analytics teams must often sift manually through these separate tools and stores to obtain insights. The process is time-consuming and arduous, and accuracy is not guaranteed.
Octopai provides an automated, centralized metadata management platform to help companies navigate their fragmented data sources: It integrates with multiple data management and storage tools, analyzes their content, and creates a consolidated repository of metadata that lets the organization pull up reports and analyses from a single location. It’s like Google but business-specific.
The same tools help in understanding the effects of scrubbing and distilling data. For example, if an organization wants to remove certain kinds of data from its stores to comply with new privacy regulations, it has to manually scan every data platform it has. And prior to making changes, it must examine how they’ll affect functionality. Automated metadata management can help a business make the necessary changes by quickly identifying relations across various platforms, and it can help the company avoid creating broken links.
“We all understand that data is the foundation of being a data-driven organization, and the critical role of metadata in all this is becoming widely recognized,” Drori says. “For so long metadata sat quietly, unharnessed, loosely managed—but today we realize that it is the key to leveraging data across every type of data process and is therefore the most important thing we have.”