Why the Majority of Big Data Projects Fail
- By Jonathan Buckley
- June 18, 2015
To truly experience growth in the future, most businesses are turning to big data. In many cases, big data is seen as the new trend guaranteed to make companies more successful. Businesses frequently turn to big data solutions for special projects designed to integrate data into normal operations and open up new business opportunities. The main problem, however, is not that these big data projects occasionally fail, it’s that they fail most of the time.
A survey from Infochimps shows that 55 percent of big data projects are never finished. This mirrors research conducted by Gartner showing a majority of data analytics projects aren’t successful. In these instances, a project fails when it isn’t completed within the set amount of budget, it isn’t finished on the established timetable, or it doesn’t have the benefits and features that were promised when the project first began. Knowing the reasons behind big data project failure can help other companies succeed while others have struggled.
Inaccurate Project Scope
One of the most glaring problems businesses face when it comes to completing a big data project is lack of proper perspective. Too often, big data projects are treated like an average IT project when it is so much more than that. Big data projects can best be considered a constantly evolving business strategy, one that needs adaptive minds and skills to deploy effectively. That strategy needs to be part of a greater vision which takes into account a company’s operational goals and aligns with them. By taking in a much wider scope, businesses will get a better view of what they’re trying to accomplish, increasing the chances of succeeding with any big data project.
Lack of Talent
Of course, to put a plan involving big data into motion, companies need to have the right people on hand with the right skill sets. This is another area where businesses often run into difficulty. Finding those with big data talents has proven to be a significant challenge. Demand for data scientists is high and the current supply is low. That turns the recruitment and hiring of skilled data scientists into a major competition among businesses with many coming up short. But it takes more than knowledge about big data to make a project successful. Businesses also need data experts with domain knowledge, or knowledge about the industry they’re in, to truly implement big data solutions that will benefit the company. Until more universities and organizations start educating and training more individuals with big data skills and industry-specific knowledge, this will continue to be a major stumbling block.
Even with the right talent on board, using big data tools can itself prove to be a challenge. Hadoop is frequently seen as synonymous with big data, yet deploying Hadoop on-premise is a sizeable task that requires the knowledge to use it. Managing on-premise Hadoop is also a tall order, especially if big data expertise is limited. In some cases, businesses simply don’t have the required internal architecture that’s needed for proper data integration. In the same Infochimps survey, 73% of those surveyed indicated understanding the big data platform was the most significant challenge faced in big data projects. When running into these formidable obstacles, it’s easy to see why so many big data projects fail.
Perhaps it shouldn’t be surprising that another major reason for big data project failure is a general lack of planning. Companies often go into a project completely underestimating just how influential and large it can turn out to be. This can lead to mismanagement of time and improper allocation of resources. Businesses may also fail to take into account how complex big data is, especially with the increasing volume and velocity inherent in big data itself. Proper planning requires skilled management and oversight along with project leaders that can clearly communicate to the whole team what is needed and how to accomplish specific tasks. Without this constant communication and management, plans often go awry and projects can quickly crumble under their own weight.
Big data project success requires a great deal of organization. Companies need to have the right talent on hand, capable of fulfilling business goals through big data solutions. They need to plan well for what the project needs while having the right tools available with the expertise to use them. For many businesses, the investment of time and resources to build and manage a Hadoop cluster and the accompanying tools is not plausible. Due to this there’s been an increasing interest in Big Data-as-a-Service. Big data in the cloud companies manage the technology and provide their own support and expertise eliminating many of the pain points listed above.