Recently, there was an intriguing article published by Chris Kanaracus on www.computerworld.com. The author quotes Ken Rudin, Facebook analytics chief, who was talking about how Big Data needs more than Hadoop.
It’s always interesting to read about the practices of successful organizations dealing with large amounts of data. In his presentation, Rudin talked about the standard topics of combining Hadoop with databases and choosing the technology suitable for the task at hand. For an analytics-minded person, these are important and sensible guidelines that should be followed.
However, what intrigued me most about this article were the insights Rudin provided regarding the analytics operation inside Facebook. It showed a full management commitment to analytics. A particularly interesting practice is Facebook’s Big Data Boot Camp program. It’s a two-week training program that teaches employees from across the organization about analytics. Having this program and learning about some of the other practices highlighted in this article make it apparent that Facebook views analytics as a core competitive advantage. What makes them successful is that they’ve acted on this insight and made a deep commitment to invest in analytics. This article confirms our belief that successful analytics must encompass a holistic approach that combines three areas of practice:
Strong data engineering discipline. Data engineering is essential for building powerful analytics. Without this discipline, it wouldn’t be feasible for companies such as Facebook to handle large amounts of data. This covers all the design areas necessary to create the right data structures best-suited for each specific technology platform. Successful organizations know how to hire the right team and invest in the right technologies and systems that will be best-suited for their environment. Choosing Hadoop or any other technology becomes a design decision based on the real needs of the organization.
Strong data science and analytics practices. Turning data into decisions requires careful analysis and preparation of the data—oftentimes involving statistical analysis. Usually, the team performing this part works closely with the business and often includes expert business users. This isn’t just about statistics; it must also include a deep understanding of the business decision-making process.
Proper management oversight and buy-in. Management has to follow the right approach to analytics. It must be viewed as an investment and an integral part of the business strategy precisely because building successful analytics isn’t an easy thing to do. Taking note of practices such as those of Facebook should drive that point home.
Management must also find creative ways to develop a culture of analytics by involving everyone in the process. Organizations fail to implement analytics successfully because they don’t recognize the principles outlined here and neglect to dedicate proper resources to each area. It isn’t enough to just build a data warehouse and declare that your analytics efforts are done. You need to couple that project with training and an ongoing improvement cycle to get the most out of it.
If you’re struggling with analytics or don’t know where to start in your organization, you aren’t alone. Tackling these areas successfully requires experience, organization and planning. Just like any other area of your business, you need to find the leaders with the right experience to make this strategy a reality. If you’re able to accomplish that task, it will allow you to take your organization beyond your competition and set you on a path to gain a strategic advantage.