As 2012 slid into 2013, several experts argued that we have to take the "big" out of big-data analytics. I agree, but not for the same reason.
The argument being made by blogs such as those on InformationWeek, consultants, and some vendors, most notably HortonWorks president Herb Cunitz, was that big-data analytics really are just about data and analytics. Rather than being something special, these analytics should just be part of the regular corporate life.
I say that we should take the "big" out because big-data is in danger of becoming a runaway freight train that could destroy a corporate life or two.
I understand where the experts are coming from. They have been watching -- sometimes guiding -- companies that are on the leading edge of analytics and have their infrastructures in place. However, a significant number of companies, perhaps a majority, don't have their big-data infrastructure in place. Looking at the SMB space, I'll bet that it's a solid majority that never even got data warehousing off the ground. While the leading edge companies are trying to figure out where they will get the staff to support big-data applications that are in production, many midsized companies are trying to figure out step one in launching analytics.
In companies of all sorts around the globe, there are marketing, logistics, human resources, product development, and quality assurance managers who have been worked into a frenzy with promises that they can tap into massive data volumes drawn from CRM systems, social media activity, and public databases just overflowing with material like census data. They think that all IT has to do is pick up one of those Hadoop thingies at Staples, and they are off and running.
Stop and take a breath. Keep your analytics goals within reach and in step with your corporate goals.
Yes, it's worth it for every large company and most midsized companies to explore big-data's potential. However, they must have a plan. Steps one and two of that plan -- evaluate your greatest business challenges and look at what types of data are available to you -- can be done in sync. Examine those two side-by-side and you will get a sense of where big-data analytics can most help your company.
Build out one, two, maybe three apps to start, and get it right. And it makes sense to utilize cloud services in most cases rather than hauling in a bunch of new hardware. Let those initial applications move you along your learning curve. Then you can take a reasoned look at where to advance your "big" data strategy with applications that might be more of a longshot in terms of providing return on investment.
If you try the shotgun approach to analytics, and try to meet everyone's new application demands right from the start, you can be sure that you will screw up the infrastructure, roll out clunky interfaces, destroy your own credibility, and end up with petabytes of data -- drawn from expensive subscriptions -- and, very possibly, some huge liabilities if you suffer a data breach.
Think "manageable" rather than "big" and you will be on the right track, not the one with the freight train rolling down from the mountain.
What advice do you have for your peers when it comes to rolling out analytics applications? Share a comment.
Related posts:
Tweet This
