"Data Driven" Requires Good Data Management
From a business perspective, big data it is about competing on analytics and making the entire organization more data driven. We explore two key points organizations seeking to become more data-driven should address.
- By David Stodder
- April 15, 2014
From a data professional's perspective, big data is about volume, velocity, variety, and some other Vs such as vim, vigor, and Vivarin (for the number of hours it usually takes to manage it all). From a business perspective, however, it is about competing on analytics and making the entire organization more data driven. Knowing that big data initiatives garner the most support when driven by specific and often immediate business needs, technology suppliers are doing all that they can to hide the complexity behind visual, easy-to-use interfaces and "data scientist in a box" packaged solutions.
The trend in big data technology marketing is to move away from talking about developer issues such as Hadoop, Python, and parallel processing and toward topics more near and dear to business executives, such as how to reduce customer churn, prevent fraud before it can do damage, or shift resources proactively to where analytics show demand is headed. Most startups that I encounter want to develop solutions that address business objectives first; only after you throw some jargon at them will they begin to discuss the innards of how their technologies dip into the big data "lake" being filled by online behavior and input from sensors, smart devices, wearable technologies, and more.
"Business-driven" business intelligence and analytics is a critically important trend; TDWI is devoting our next World Conference to the topic and it is the subject of the next Best Practices Report I will be writing. Most business users, who are subject matter experts in their own areas of concern, have little time or inclination to interrupt their focus on business needs to learn technology details. Some still resist using BI and analytics tools, preferring to do their own analysis on spreadsheets or just go with gut feel and intuition. If they are to become more data-driven, most business users want to stay at a higher level. Thus, to appeal to the business side, the technology mumbo jumbo must be kept to a minimum.
Flying Data Blind
Organizations could be in danger if they convince themselves that by merely installing packaged solutions or subscribing to online services they will have become "data-driven" and ready to conquer the world with analytics. Even startups that seek to base their competitive advantage on data science can be vulnerable to the illusion that the complexities of underlying data management and governance can be preconfigured away and hidden behind slick visualizations. From what I've seen, few data scientists have an adequate understanding of, or even interest in, these topics.
Here are two key points that organizations seeking to become more data-driven should address:
#1: Know more about your data. Unless organizations know where their data is located and how it is being used, they will fall short of creating the strategic asset that they envision. The chances are too high that as users try to perform analytics across sources, data quality, profiling, and metadata errors will increase the "noise" level so high that the effort will grind to a halt in confusion. Organizations should take steps to discover how data elements are related within and across data stores; it would be best to document what they find so that have a record of valid data and schema relationships in a metadata repository, business glossary, or similar system.
#2: Take data governance seriously. Pressures are rising on organizations to take better care of customer data and adhere to regulations governing the use of information. Governance policy enforcement involves identifying, securing, and managing data in multiple systems at multiple locations. Without reliable and repeatable methods of understanding where data is and how it is being transformed, organizations will struggle to respond to current regulatory obligations and address future ones.
Being Data-Driven Requires Management
With marketing, finance, and other department and line-of-business managers commanding technology budgets for BI and analytics, it's clear that the "business-driven BI and analytics" trend is here to stay. IT managers have to adjust; enterprise BI and data warehousing systems need to accommodate different data demands. As they pursue dreams of competing on analytics, organizations must take good care of what makes it all possible: the data.