The buzz around Big Data is unmistakable. And around the world, startups and corporate giants alike are leveraging Big Data and Big Data applications to create value—quite literally, turning terabytes of transactions into hard cash and competitive edge.
But behind the gloss, there’s an inconvenient truth. Simply put, not all of those attempts to wrest value from Big Data will be successful. Far from it.
And here’s another inconvenient truth. In the world of enterprise IT, ‘success’ with Big Data has a somewhat elastic definition, one that sometimes sounds like “At least we didn’t completely fail.”
So how can you tell if a given Big Data application is a genuine success? Here are four signs that distinguish a true Big Data success from an implementation that didn’t simply avoid a crash landing.
1. It works
Big data creates concrete value across industries, not limited to high-tech. McKinsey’s report on the future of Big Data identified over a trillion dollars of potential value gains from big data in the health care, government, retail, and manufacturing sectors. A successful implementation within your organization should be measurable in additional revenue, improved customer satisfaction, reduced costs, or some other criterion critical to your success.
2. It delivers disruptive change
Big data should deliver more value than mere incremental improvements of existing business models. Take the startup Foursquare as an example. To uncover important relationships among its data, Foursquare applied machine learning algorithms that led it build ‘Explore’, a social-recommendation engine that provides users with valuable venue suggestions in real-time and drives new foot traffic to new businesses. ‘Explore’ relies on big data hubs that draw insights from more than 30 million venues simultaneously. Now Foursquare has the ability to understand how people interact with each other and venues not just on the platform, but in the ‘real’ world.
3. It’s fast
Tackling Big Data with legacy database technologies is slow and cumbersome because proprietary licensing involves corporate bureaucracies long before you know whether the technology will even meet your needs. A successful Big Data project, using toolsets and database technology designed from the ground up to work with the diversity and volume of Big Data, redefines the art of the possible. The proof: A Hadoop cluster can be set up in a matter of hours, and provide useful analytics quickly after. The fact that most big data technologies are open source means that you can add support and services as you need them; licensing isn’t an obstacle to speedy implementation.
4. It delivers applications that weren’t possible before
Before the advent of Big Data technologies, "flash sales" companies like Gilt Groupe couldn’t exist. Flash sales sites need to be able to handle millions of users logging on each day for exclusive deals, resulting in extremely spiky server loads – a business model that becomes possible with high-performance, quickly-scalable Big Data technologies.
And the key lesson from all this? Trying to achieve Big Data success with outdated, legacy data infrastructure is like Jimmy Connor trying to win Wimbledon in 2013: it’s possible but not likely. Connor was a giant in his day, but the game (and his body) has moved on. While great for yesterday’s systems of record, legacy database technologies are a poor fit for fast, flexible Big Data applications that deliver disruptive changes to your business. To succeed in modern Big Data challenges, you need modern data infrastructure like NoSQL databases and Hadoop.