Five Dos and Don’ts for Managing Data Assets

Register now

More and more companies are investing in analytics platforms to boost productivity and achieve – and maintain – a competitive advantage. After implementation, it is important to understand how to leverage the full capabilities of analytics, and avoid some data management “bad habits” that could be hindering the efficiency of your organization.

I’ve observed a few data management best practices to adopt, and blunders to avoid, when managing information and data assets. So, to help organizations weigh the right criteria when implementing their analytics strategies, I’ve developed five dos and don’ts for managing information and data assets when working closely with customers:

1. Don’t forget about veracity – People tend to remember volume, variety and velocity, but neglecting veracity, or the validity of your results, can prevent an honest assessment of your predictive model. When measuring the validity of a particular dataset, no saying rings more true than “one squeaky wheel gets all the attention.” As data scientists, our first instinct is to think about the big picture conclusions gathered from a particular dataset to suggest conclusions that lead to better decision-making. However, starting small and measuring the veracity of a dataset is the first step in avoiding the inaccurate insights.

2. Do manage the metadata – Although it’s challenging to manage, metadata provides information about your existing data to help people organize how they want to use it moving forward. As with veracity, paying attention to the metadata gives data scientists an additional proof point when defending their assertions. Why should organizations be more cognizant of the data about the data? Metadata is important as you consider volume: storage for data is so affordable that organizations are often storing – in perpetuity – more data than they know what to do with. Metadata offers a more targeted analysis to help the organization better understand how to use that data.

3. Do take advantage of internal resources – For many organizations, the thought of adding a data scientist to the team is daunting. Companies are scaling back, budgets are being cut and the need to be resourceful with your current team structure is critical. Before you fret, take a closer look at the talent within your organization. Often, organizations already have the expertise in-house. Data scientists are often mathematicians, statisticians and data analysts who are skilled in seeing things in a business-centric way. If you can identify someone in your organization with an analytical mindset and the ability to provide recommendations to drive change, empower them to play the role. This is the era of the “citizen analyst,” after all.

4. Do combine the old with the new – Combining new big data analytics with the old traditional business intelligence will help you gain the hindsight needed to create momentum in new opportunities and trends. Speaking in terms of infrastructure, if companies want to break down silos and make returns on their investments, it is vital for their IT departments to pair new Hadoop clusters with existing data warehouses. On the data analytics side, taking a closer look at how predictive models were run in the past is crucial to laying the foundation for understanding the data and figuring out where to potentially reuse big picture trends.

5. Don’t neglect security – It’s easy to overlook data security as a key business enabler, but with the number of high-scale security breaches on the rise, the topic is becoming harder to ignore. As companies collect and store more and more data, it is paramount to ensure your customers’ information is protected. The rise of the Internet of Things and connected devices introduces new security vulnerabilities to the industry. Connected cars, homes and other devices connecting to the internet can become an entrance for attackers into the enterprise. Perhaps the most important rule of all is to ensure that your organization knows how to handle the security risks involved with data management. With the new sources of data becoming more important to a complete analytic view, the combination of “secure” data with “publicly” available data needs to be meticulously managed. The management can be done so long as organizations implement transparent and conscientious protocols or processes. Security is an issue that is solved with both technology and business process – you need an equal share of both to be successful and to safeguard your business from data breaches.

If you want to become a valuable resource on your data science team, it’s important to weight the dos against the don’ts. In summary, pay attention to the validity of your data, closely analyze the metadata (the data about your data), take advantage of the resources you have available inside your organization, combine new analytics with traditional business intelligence, and be aware of the security vulnerabilities.

Joanna Schloss is a business intelligence and analytics expert at Dell Software.

For reprint and licensing requests for this article, click here.